Multidimensional analysis of supervisors’ safety leadership on safety violations of construction workers: An empirical investigation

Abstract

BACKGROUND:

Frontline supervisors have the most frequent interactions with workers on construction projects. Although Supervisors’ Safety Leadership (SSL) is commonly practiced, its specific inter-relationship with workers’ safety violations remains unclear, especially when it comes to detailed interactions between supervisors and workers, such as supervisors’ safety coaching/safety controlling/safety caring against workers’ situational/routine safety violations.

OBJECTIVE:

This study aims to uncover the intrinsic relationship between SSL and safety violations from the perspective of construction workers with the help of mediating variables at both organizational and individual levels.

METHODS:

A questionnaire survey was conducted to test all hypotheses based on empirical data from 346 construction workers. The path coefficient of the fitted model was then analyzed, including associated mediating effects.

RESULTS:

Situational safety violations are directly affected only by safety caring (β= –0.161, p < 0.05), while routine safety violations are impacted only by safety coaching (β= –0.159, p < 0.05). SSL can influence different types of safety violations through differing mediators. In particular, safety coaching acts on individuals’ routine safety violations mainly through self-efficacy (β= 0.199, p < 0.01; standardized indirect effect = –0.121, 95% CI[–0.226, –0.024]); safety controlling is more oriented to influence individuals’ situational safety violations through group safety norm (β= 0.383, p < 0.001; standardized indirect effect = –0.091, 95% CI[–0.177, –0.036]); and safety caring further influences individuals’ situational safety violations mainly through safety motivation (β= 0.581, p < 0.001; standardized indirect effect = –0.263, 95% CI[–0.418, –0.146]).

CONCLUSION:

The research enhances existing knowledge by clarifying the complex relationships between supervisor behavior and safety outcomes, particularly from the perceptions of construction workers towards supervisors’ actions and leadership.

Keywords

Supervisors’ safety leadership construction worker safety violations safety motivation group safety norm self-efficacy

1 Introduction

Construction accidents lead to around 60,000 fatalities annually worldwide, equivalent to one accident every nine minutes [1, 2]. The incidence of non-fatal injuries in the construction industry is nearly 30% higher than in other industries [3, 4], indicating a high incidence of both fatal and non-fatal injuries in this industry. The frequent and severe consequences of construction accidents (e.g., casualties) have led to continuous exploration of their causes. Workers’ safety violations are identified as one of the major contributors to construction accidents [5 –7]. Suraji et al. [8], in their analysis of 500 construction accidents in the United Kingdom, attributed approximately 80% of these accidents to workers’ safety violations. Hinze and Rinker [9] also stated that more than 75% of construction accidents were caused by safety violations among the workforce. These violations are influenced not only by the work environment and tasks but also by the behavior of other staff in the same organization, especially managerial behavior [10 –14]. Specifically, when managers demonstrate strong safety awareness and behaviors and effectively communicate them to workers, it significantly mitigates workers’ safety violations [7]. For managers, the ability to ensure safety and guide their subordinates towards common safety goals has been defined as “Safety Leadership” [15, 16]. Safety leadership provides an innovative research direction for the exploration of the safety behavior of workers, and some empirical studies have investigated its effectiveness in accident rate reduction [5 , 17–19].

With the introduction of safety leadership, increasing attention has been paid to its impacts on workers’ safety violations. Due to the multi-layered structure of organizations and the varying roles and responsibilities of managers in the construction industry, there are different forms of safety leadership, each with varied influences on workers’ safety violations. Flin and Yule [20] divided the management structure of construction projects into three layers: the strategic layer, the tactical layer, and the frontline operational layer. The strategic layer generally comprises a senior management team led by a project manager [21]. The tactical layer usually contains middle managers (e.g., department heads of a construction project) [21]. The frontline operational layer consists mainly of the junior managers of construction projects, also known as “supervisors” [7 , 23]. Supervisors have direct access to the frontline workers and supervise their daily production and operation behavior [21 , 24]. Compared with middle and senior managers, supervisors play more critical roles in accident prevention, given their closest relationships and most frequent interactions with workers [21 , 25–27]. However, the inter-relationships between the safety leadership of supervisors and workers’ safety violations remain unclear, particularly regarding the specific relationship between the different dimensions contained in both [7].

Based on the above description, this study aims to reveal the intrinsic relationship between SSL and safety violations from the perspective of construction workers. It intends to bridge the research gap in related theoretical systems and management practices from three aspects, in order to make more in-depth theoretical contributions and practical value. First, although existing literature has provided insights into the relationship between supervisors’ safety leadership and workers’ safety violations (or compliance) [28, 29], many studies have focused solely on SSL’s impact on safety violations [30 –32]. This study, however, more comprehensively considers the three core dimensions of SSL: safety coaching and safety caring, characteristic of transformational leadership, and safety controlling, characteristic of transactional leadership [33, 34]. In addition, it incorporates two dimensions of safety violations: individuals’ routine safety violations and individuals’ situational safety violations [7 , 35–37]. This multidimensional division not only demonstrates the varying impact of different types of SSL but also helps to explore the mechanisms of the two types of safety performance (IRSV and ISSV). Second, prior research on the impact of SSL on safety violations has often focused on a single mediating variable at the organizational level [28 , 38], which leaves the complex path relationship between SSL and the multiple dimensions of safety violations unexplained. In this regard, by referring to Self-Determination Theory, Social Cognitive Theory and Social Information Processing Theory, we consider three mediating variables— group safety norms at the organizational level, safety motivation of workers at the individual level, and self-efficacy— drawing from Self-Determination Theory, Social Cognitive Theory, and Social Information Processing Theory. These variables are examined to reveal the mechanisms mediating the relationship between SSL and safety violations. Finally, it is expected that the results of the study will help management to target the improvement of different types of safety performance among construction workers through different types of SSL in practice. Meanwhile, this is also expected to contribute to the sustainable development of safety management in construction workplaces.

The remainder of this paper is structured as follows. The next part comprises the literature review, which lays the groundwork for research hypotheses through a comparative and analytical examination of current research. The third part presents information on the data collection and research design, followed by the data analysis and results. After the discussion of the findings, the conclusion, the limitations and perspectives for future research are presented.

2 Literature review

2.1 Safety violations

Construction safety violations are the opposite of safety compliance, referring to behaviors intentionally carried out by construction workers during the operation process, which, although unintentionally causing any harm, violate the organizational safety system and procedures [7 , 40]. Safety violations are usually categorized into two dimensions: individuals’ routine safety violations (IRSV) and individuals’ situational safety violations (ISSV). IRSV refer to the intentional circumvention of safety systems, procedures, or practices by workers to achieve organizational (e.g., timely completion of work tasks) or personal (e.g., saving time or physical effort) benefits. When committing routine safety violations, workers habitually choose the most physically efficient operation or a “shortcut” [41]. In contrast, IRSV and ISSV result from situational constraints in the workflow that render compliance with the safety system impractical [42]. For instance, an operator may violate a safety system when protective equipment is unavailable. The main difference between these types of safety violations lies in whether they are caused by situational constraints. Existing studies indicate that situational and routine safety violations are influenced by different antecedent variables. For example, Liang et al. [39] found that workers’ perceived social support and safety motivation affect ISSV, while perceived production stress and ambivalent attitudes affect IRSV. Another study by Liang and Zhang [40] concluded that workers with low education levels are more likely to violate situational safety regulations, while inexperienced workers are more inclined towards routine safety violations. On this basis, this study also classified construction workers’ safety violations into IRSV and ISSV. It explored the intrinsic mechanisms between the different dimensions of supervisor safety leadership and these two types of safety violations.

2.2 Supervisors’ safety leadership (SSL)

2.2.1 Definition of SSL

Existing research has conceptualized “safety leadership”. For instance, Wu [33] defined safety leadership as “the interactive process by which leaders influence their followers to achieve organizational safety goals in the context of organizational and individual factors”, a definition widely accepted. Based on this perspective, safety leadership is a process, a leader’s influence on subordinates in group contexts and aimed at shared safety goals. Numerous studies have advocated that supervisors be placed at the center of safety interventions for construction workers [21 , 43]. Despite lower positions in the management structure and the least authority, supervisors have the most frequent and close contact with workers and play a crucial role in construction safety management [7]. Supervisors enforce virtually all safety-related policies and regulations, coordinating subordinate work assignments. They oversee on-site operational activities and serve as intermediaries for communicating workers’ safety concerns, questions, and requests to top management [21]. In conjunction with these responsibilities, this paper defines the concept of “supervisors’ safety leadership (SSL)” in construction projects as “the safety competencies and means that supervisors possess during the construction process, which are able to influence construction workers and guide them to complete their production tasks safely”.

2.2.2 Dimensions of SSL

In terms of the dimensions of SSL, this study referred to the dimensions defined by Wu [16] and Wu [33]. According to these two studies, the SSL should consider characteristics of transformational leadership and transactional leadership, as the paradigms of transactional and transformational leadership are comprehensive enough to measure and understand leadership structures. However, before the analysis of these dimensions, it is crucial to have a certain understanding of transformational leadership and transactional leadership.

Transformational leadership emphasizes change [5]. It is designed to enable employees to clearly understand their responsibilities and obligations through flexible and varied leadership styles, thereby stimulating a higher level of cognition. Consequently, employees can achieve higher levels of performance by tapping into their potential to a greater extent [5]. In the context of construction workers’ safety performance, transformational leadership influences through four dimensions: idealized influence or charisma (behavioral role model), visionary motivation (developing subordinates’ commitment to goals), intellectual stimulation (helping subordinates solve problems intellectually and operationally), and personalized care (understanding and empathizing with subordinates) [29]. On the other hand, Barling et al. [15] systematically outlined four distinguishing characteristics of transactional leadership, including clear hierarchical relationships, good order, attention to rules, and obsessive control.

Transformational leadership behavior favors employee-centeredness, promoting performance by coaching employees on skills and ideas while expressing concern and encouragement. In contrast, transactional leadership tends to be more manager-centered, emphasizes clear work-management hierarchy boundaries, and promotes a controlled work environment through a forceful leadership style. On the basis of the two leadership styles and the characteristics of supervisors in the construction industry, this study identified three dimensions of SSL: Safety Coaching (SCoa), Safety Controlling (SCon), and Safety Caring (SCar). Safety coaching and safety caring align with aspects of transformational leadership; safety controlling is closely linked to transactional leadership [33]. These dimensions are defined as follows:

SCoa means that supervisors can be a role model for workers to learn from, can provide guidance on daily operations; can mobilize their intellect; can share ideas; and can involve workers in decision-making [33, 34].

SCon refers to the extent to which supervisors set standards of behavior for workers according to a system of rules; use authority to correct violations; and use technology to monitor safety performance [33, 34].

SCar includes considering problems from the perspective of construction workers, listening to their safety feedback patiently, understanding their safety requirements, considering their safety concerns, and providing timely safety support and help in action [33, 34].

2.3 SSL and safety violations

SSL has been shown to be highly effective in improving human safety behavioral performance. For example, Dickerson et al. [44] argued that behavior-based safety coaching is a successful strategy for reducing errors in the context of industrial occupational safety. Leader coaching behavior can effectively guide subordinate safety behavior [45]. Kapp [28] and Petitta et al. [46] both pointed out that strong safety control can directly enhance workers’ safety performance and reduce the frequency of their mistakes. Liu et al. [47] found that safety support from superiors can directly and positively influence employees’ safety citizenship behavior. Based on these findings, we argue that SSL can attenuate worker safety violations and propose hypotheses 1 and 2.

H1: Three sub-dimensions of SSL, namely Safety coaching (H1a), Safety controlling (H1b), and Safety caring (H1c), have negative influences on ISSV.

H2: Three sub-dimensions of SSL, namely Safety coaching (H2a), Safety controlling (H2b), and Safety caring (H2c), have negative influences on IRSV.

2.4 Safety motivation as the mediating variable

Safety motivation (SM) refers to an individual’s willingness to make efforts to achieve safety and the corresponding actions, reflecting the importance employees attribute to safety [48 –50]. Previous research conceptualized safety behavior as an interaction between proximal individual differences (e.g., safety motivation, etc) and distal situational factors (e.g., leadership or safety management, etc). Therefore, safety leadership has been recognized as an important antecedent of safety motivation. Current research has confirmed this relationship. For example, Kovjanic et al. [51] argued that when supervisors provide careful instruction and safety care, these high-quality behaviors resonate with workers and enhance their sense of identification with the organization, which helps workers foster safety motivation. Probst and Brubaker [52] noted that workers’ safety motivation relates to supervisors’ perceptions of safety policy enforcement, including the extent to which supervisors commend safety compliance and sanction non-compliance. Inadequate enforcement by supervisors correlates with low levels of worker safety motivation. Based on the above discussions, hypothesis 3 is proposed:

H3: Three sub-dimensions of SSL, namely: Safety coaching (H3a), Safety controlling (H3b), and Safety caring (H3c) have positive influences on SM

In addition, existing research generally recognizes safety motivation as a key determinant of whether an individual makes safety violations [39]. Safety motivation can significantly enhance employees’ level of safety compliance [53 –55]. However, the extent to which safety motivation plays a role in safety violations may vary depending on the nature of the violations. For example, when workers still choose to adhere to safety rules without safety protective equipment, they may need more safety motivation. On the contrary, if they are adhered to just to comply with safety procedures, the effectiveness of safety motivation may be relatively low [48]. Therefore, hypothesis 4 is proposed to test the effect of safety motivation on different types of safety violations:

H4: SM have negative influences on two sub-dimensions of safety violations, namely: ISSV (H4a) and IRSV (H4b)

On the basis of hypotheses 3 and 4, we further combine the Self-Determination Theory (SDT) to explain the mediating role of safety motivation between SSL and safety violations: SDT recognizes human beings as positive organisms with innate potential for psychological growth and development [56]. However, these intrinsic potentials often require external stimuli for reinforcement and gradually develop into control over behavior. Therefore, key external environmental variables (e.g., leadership) help stimulate autonomous motivation in employees to further drive their growth and progress [51]. Therefore, hypothesis 5 is proposed:

H5: SM can mediate the relationship between SSL and safety violations: SM can mediate the relationship between the three sub-dimensions of SSL (Safety coaching (H5a), Safety controlling (H5b), and Safety caring (H5c)) and ISSV. SM can mediate the relationship between the three sub-dimensions of SSL (Safety coaching (H5d), Safety controlling (H5e), and Safety caring (H5f)) andIRSV.

2.5 Group safety norm as the mediating variable

Group safety norm (GSN) indicates the social pressure to perform safety behavior, namely, what others expect the individual should undertake to perform tasks safely. It is a collective cognition of a group [57]. The common belief of group is formed by the interaction between internal members. The interaction occurs not only between workers, but also between workers and their direct leaders [21, 58]. Previous studies have affirmed the pivotal role of supervisors and their leadership in the formation of safety-related group norms in construction teams. For example, Fang et al. [21] found that the training and prevention behavior performed by supervisors significantly affects the group-level safety climate, including the involvement and mutual influence of workers. McFadden et al. [59] argued that reasonable control over employees can effectively cultivate positive group safety norms. He et al. [60] stated that a high-quality leader-member exchange relationship directly and positively affects the safety climate of the team. Based on the above discussions, hypothesis 6 is proposed:

H6: Three sub-dimensions of SSL, namely: Safety coaching (H6a), Safety controlling (H6b), and Safety caring (H6c) have positive influences on GSN

Moreover, due to the complexity of the construction site, formal rules cannot regulate every aspect of workers’ behavior at all. Behavior changes driven by social norms are more sustainable and effective, especially when such social norms spread to the work groups [61]. Under such an environment, the safety behavior of workers can be improved unconsciously, and they will do their best to avoid violations. Therefore, hypothesis 7 is proposed:

H7: GSN have negative influences on two sub-dimensions of safety violations, namely: ISSV (H7a) and IRSV (H7b)

Based on hypotheses 6 and 7, we further combine the Social Cognition Theory to explain the mediating role of group safety norm between SSL and safety violations: according to social cognition theory, individual workers always try to belong to a social group and adjust their behavior to what they perceive as socially acceptable within the crew [62]. Despite the belief in the effectiveness of GSN in mitigating the relationship between SSL and worker safety violations, few studies have synthesized the multidimensionality of SSL and safety violations. Therefore, hypothesis 8 is proposed:

H8: GSN can mediate the relationship between SSL and safety violations: GSN can mediate the relationship between the three sub-dimensions of SSL (Safety coaching (H8a), Safety controlling (H8b), and Safety caring (H8c)) and ISSV. GSN can mediate the relationship between the three sub-dimensions of SSL (Safety coaching (H8d), Safety controlling (H8e), and Safety caring (H8f)) and IRSV.

2.6 Self-efficacy as the mediating variable

In the field of safety, workers’ self-efficacy refers to “workers’ estimates and judgments of their ability to accomplish safety-related goals and tasks” [63]. Evidence has shown that in safety critical industries, the involvement of supervisors in safety management can effectively enhance the self-efficacy of subordinates towards safety rules. Similarly, the positive relationship between SSL and employees’ self-efficacy has been widely demonstrated. Bono and Ilies [64] suggested that subordinates emulate the exemplary behavior of their supervisors to stimulate their own positive feelings and efficacy. Fugas et al. [65] argued that supervisors’ “injunctive norms” contribute to workers’ accurate perceptions of their own abilities and motivate them to surmount difficulties. In addition, care and support from supervisors have been identified as effective drivers of self-efficacy [66]. However, few studies have comprehensively compared the differences in the impact of different dimensions of SSL on self-efficacy. Based on the above discussions, hypothesis 9 is proposed:

H9: Three sub-dimensions of SSL, namely: Safety coaching (H9a), Safety controlling (H9b), and Safety caring (H9c) have positive influences on SE

Additionally, several studies have reported a positive relationship between workers’ self-efficacy on safety performance. For instance, Yen et al. [67] and Saleem et al. [68] both concluded that workers’ self-efficacy positively affects their safety compliance. Kim and Jang [69] revealed that safety self-efficacy is a key skill for individuals to inhibit human errors and thus avoid potential accidents. Therefore, we believe that the self-efficacy of workers can reduce the probability of safety violations and thus propose hypothesis 10:

H10: SE have negative influences on two sub-dimensions of safety violations, namely: ISSV (H10a) and IRSV (H10b)

Based on hypotheses 9 and 10, we further combine the Social Information Processing (SIP) theory to explain the mediating role of self-efficacy between SSL and safety violations: according to SIP, an individual processes information based on their understanding of how cognitive and behavioral responses come about through social interactions [70, 71]. Since behaviors, attitudes, and perceptions are affected by social information, all aspects of individual behaviors, attitudes, and perceptions are presented as a result of cognitive products of information processing [72, 73]. In construction workplaces, workers heavily rely on cues or signals from leaders to confirm their understanding of the environment in the organization and adjust their cognitions accordingly. Thus, in explaining the link between safety leadership and safety violations, we hypothesized a mechanism that could potentially mediate this relationship. From this perspective, workers’ self-efficacy was identified as another mediator at the individual level. Therefore, hypothesis 11 is proposed:

H11: SE can mediate the relationship between SSL and safety violations. SE can mediate the relationship between the three sub-dimensions of SSL (safety coaching (H11a), safety controlling (H11b), and safety caring (H11c)) and ISSV. SE can also mediate the relationship between the three sub-dimensions of SSL (Safety coaching (H11d), Safety controlling (H11e), and Safety caring (H11f)) and IRSV.

Based on the above analysis, the complete research model of this paper is shown in Fig. 1.

Fig. 1

The hypothesized model regarding the supervisors’ safety leadership-safety violations relationship.

3 Materials and methods

3.1 Participants and procedures

The questionnaire survey was conducted between 01/05/2023 and 31/08/2023. Data were collected using both electronic and paper questionnaires. A valid survey link was provided through the questionnaire website (Question-star) for respondents to fill in the questionnaire. As per the research scope from the perspective of construction workers, 500 paper questionnaires were distributed to frontline construction workers in major Chinese cities, including Shanghai, Suzhou, Nanjing, Chongqing, Xi’an and Wuhan. We collected 255 paper questionnaires and 198 online questionnaires. After excluding those that did not meet the requirements, 195 valid paper questionnaires and 151 online questionnaires were finally recovered, with valid questionnaire recovery rates of 76.5% and 76.3%, respectively. A total of 346 valid questionnaires were recovered. The demographic information of the respondents is shown in Fig. 2. These sample characteristics are consistent with the nature and uniqueness of the construction industry.

Fig. 2

Demographic information of the respondents. Note: ^*represents p < 0.05, ^**represents p < 0.01.

3.2 Questionnaire development and measurement

We adapted previously validated measurement scales and developed specific scales for use in the construction context. Before the formal scale was finalized, the original English questions were translated into Chinese following standard translation procedures, followed by a panel discussion among five experts in the field of engineering safety to determine the final measurement scales (e.g., incorporating industry experience and knowledge). Each measurement question was rated on a Likert 5-point scale ranging from 1 (strongly disagree) to 5 (strongly agree). Table 1 shows the variables, measures of study variables, number of items, and supporting literature.

Table 1
Variables, measures of the studied variables, number of items, and supporting literature

Variables Measures of the studied variables Items Supporting literature

Supervisors’ safety leadership (SSL) Safety coaching (SCoa) 6 Wu [33]

Safety controlling (SCon) 7

Safety Caring (SCar) 6

Safety motivation (SM) 3 Nykanen et al. [74]

Group safety norm (GSN) 5 Su et al. [7]

Self-efficacy (SE) 6 Luthans et al. [75]

Safety violations (SV) Individuals’ routine safety violations (IRSV) 4 Chmiel et al. [35]; Su et al. [7]

Individuals’ situational safety violations (ISSV) 6

Variables	Measures of the studied variables	Items	Supporting literature
Supervisors’ safety leadership (SSL)	Safety coaching (SCoa)	6	Wu [33]
	Safety controlling (SCon)	7
	Safety Caring (SCar)	6
	Safety motivation (SM)	3	Nykanen et al. [74]
	Group safety norm (GSN)	5	Su et al. [7]
	Self-efficacy (SE)	6	Luthans et al. [75]
Safety violations (SV)	Individuals’ routine safety violations (IRSV)	4	Chmiel et al. [35]; Su et al. [7]
	Individuals’ situational safety violations (ISSV)	6

A 19-item scale was used in this study to assess each worker’s perceptions of safety leadership across three adapted subscales: Safety coaching, Safety controlling and Safety caring [33]. Sample question items include, “My supervisor helps workers realize the importance of safety” (SCoa), “My supervisor orders us to be determined to accomplish safety goals” (SCon), “My supervisor works hard to meet our safety needs” (SCar), etc.

The scale used to measure worker safety motivation was selected from the validated scale developed by Nykanen et al. [74], which measured safety motivation among vocational high school students with three items. We fine-tuned its content by incorporating the context of the construction safety industry (one of the sample items was “I feel that working to maintain or improve my personal safety is worthwhile”).

To measure worker group safety norms, we adapted a scale from the work of Su et al. [7]. This scale contains five items to assess workers’ perceptions of whether fellow team members recognize certain behaviors. One of the sample items is: “Members within our shift would strongly disapprove of behavior that violates safety rules and regulations”.

The scale used to measure workers’ self-efficacy was selected from the Psychological Capital Scale developed and validated by Luthans et al. [75], which measured employees’ self-efficacy using six items. We adjusted it to incorporate the context of the construction safety industry (one of the sample items was: “I believe that I can always analyze problems and find solutions independently”).

The questions for assessing both routine and situational safety violations were adapted from the self-report items validated by Chmiel et al. [35] and Su et al. [7]. Routine safety violations were measured with four items, with a high score indicating a greater frequency of violations. A sample item reads, “I’ll take shortcuts to get the job done if I can make it easy”. Situational safety violations were measured with six items and were reversely scored: a high score indicated a low level of violations. A sample item is “Even if it’s inconvenient sometimes, I always wear a safety helmet”.

3.3 Data analysis

The data analysis was conducted in three stages using SPSS 21.0 and AMOS 24.0 software. First, SPSS 21.0 was used for descriptive analysis. Second, reliability tests were conducted using SPSS 21.0, and AMOS 24.0 was used for confirmatory factor analysis (CFA) to test the dimensionality and discriminant validity of our multiple-item measures, and metrics were calculated to test convergent validity. Finally, all hypotheses (across the entire model) were tested (see Chapter 4), including tests for direct effects between variables as well as mediating effects (using AMOS 24.0 software).

3.3.1 Descriptive analyses

Table 2 shows the results of the descriptive statistical analysis. The smaller standard deviation indicates a lesser degree of variation in the data, which means that the participant’s level of understanding and answers to the questions were at an acceptable level. In addition, the absolute values of kurtosis for each variable are less than 7 and the absolute values of skewness are less than 3. These observations suggest that the variables follow a normal distribution and are suitable for analysis [76 –78].

Table 2
Results of the descriptive statistical analysis

Variable Name Mean SD Kurtosis Skewness

SCoa 3.891 .862 .878 –1.090

SCon 3.628 .913 –.324 –.702

SCar 3.837 .912 .920 –1.249

SM 3.967 .946 1.306 –1.291

GSN 3.853 .915 1.286 –1.263

SE 3.518 1.130 –.744 –.708

ISSV 2.219 .865 .280 .966

IRSV 3.090 1.135 –1.263 –.172

Variable Name	Mean	SD	Kurtosis	Skewness
SCoa	3.891	.862	.878	–1.090
SCon	3.628	.913	–.324	–.702
SCar	3.837	.912	.920	–1.249
SM	3.967	.946	1.306	–1.291
GSN	3.853	.915	1.286	–1.263
SE	3.518	1.130	–.744	–.708
ISSV	2.219	.865	.280	.966
IRSV	3.090	1.135	–1.263	–.172

3.3.2 Reliability and validity testing

The internal consistency of the variables underwent verification through a reliability test. Cronbach’s α coefficient was used to quantify reliability with a lower threshold value of 0.6–0.7 [79]. It can be seen that the Cronbach’s α values of the variables are greater than 0.8 (shown in Table 3), indicating strong reliability.

Table 3
Results of the variables’ convergent validity

Variable Indicator SFL AVE CR α

Safety coaching SCoa1 0.823 0.683 0.928 0.926

SCoa2 0.858

SCoa3 0.829

SCoa4 0.742

SCoa5 0.830

SCoa6 0.871

Safety controlling SCon1 0.789 0.629 0.922 0.922

SCon2 0.797

SCon3 0.806

SCon4 0.807

SCon5 0.844

SCon6 0.733

SCon7 0.770

Safety caring SCar1 0.811 0.714 0.937 0.937

SCar2 0.804

SCar3 0.769

SCar4 0.951

SCar5 0.797

SCar6 0.920

Safety motivation SM1 0.856 0.695 0.872 0.872

SM2 0.851

SM3 0.793

Group safety norm GSN1 0.822 0.700 0.921 0.920

GSN2 0.873

GSN3 0.811

GSN4 0.822

GSN5 0.855

Self-efficacy SE1 0.852 0.749 0.947 0.947

SE2 0.839

SE3 0.912

SE4 0.852

SE5 0.851

SE6 0.885

Individuals’ routine safety violations IRSV1 0.840 0.728 0.914 0.913

IRSV2 0.901

IRSV3 0.815

IRSV4 0.854

Individuals’ situational safety violations ISSV1 0.783 0.652 0.918 0.915

ISSV2 0.777

ISSV3 0.848

ISSV4 0.785

ISSV5 0.821

ISSV6 0.829

Confirmatory factor analysis (CFA) was used to test the validity of the measurement model. The validity test usually consists of two parts: the convergent validity test and the discriminant validity test. Convergent validity assesses the extent to which the latent variables can be measured and explained by their observed variables. It is assessed using three metrics: standardized factor loadings (SFL), combined reliability (CR), and average variance extracted (AVE), with recommended thresholds of 0.5, 0.7, and 0.5, respectively. The results show the SFLs are all above 0.5 [80], while the CR and AVE values are above 0.7 and 0.5, respectively [81] (shown in Table 3). Therefore, the convergent validity is deemed acceptable.

Discriminant validity assesses the degree to which a latent variable differs from other latent variables and can be measured by comparing the square root of the variable AVE with its maximum correlation coefficient. When the former is greater than the latter, the variables are considered to have good discriminant validity [82]. For instance, the square root value of AVE for SCoa is 0.826, greater than its maximum correlation coefficient with SM of 0.442. This indicates that SCoa demonstrates acceptable discriminant validity. Similarly, as shown in Table 4, the other seven variables also show good discriminant validity.

Table 4

Discriminant validity test

	1	2	3	4	5	6	7	8
SCoa	(.826)
SCon	.394^**	(.793)
SCar	.395^**	.329^**	(.845)
SM	.442^**	.448^**	.681^**	(.834)
GSN	.354^**	.501^**	.504^**	.509^**	(.837)
SE	.192^**	.066^**	.085^**	.075^**	.124^**	(.865)
ISSV	–.349^**	–.462^**	–.611^**	–.659^**	–.561^**	–.080^**	(.807)
IRSV	–.303^**	–.213^**	–.135^**	–.207^**	–.213^**	–.448^**	–.267^**	(.853)

^**p < 0.01. Note: diagonal values are the square roots of average variance extracted (AVE).

4 Results

After analyzing the reliability and validity of the variables, the hypotheses in the model require further testing. Our hypotheses contain direct paths as well as mediated paths between the variables, so both types of relationships need to be tested in turn.

4.1 Structural model fit

To clarify the direct influence relationship among the variables, we analyze the model composed of all variables using AMOS 24.0 software. The first step is to determine the validity of the structural equation model [83], mainly gauged through the measurement of some fit indicators [84, 85]. The model fit metrics and their corresponding criteria are shown in Table 5. All metrics meet the criteria except for the NFI, which is close to 0.9. This suggests an acceptable overall model fit.

Table 5
Model goodness-of-fit index

Fit Metrics Acceptable range Measurements

χ²/df <3 1.592

RMSEA <0.08 0.041

IFI >0.9 0.959

NFI >0.9 0.897

TLI >0.9 0.956

CFI >0.9 0.959

Fit Metrics	Acceptable range	Measurements
χ²/df	<3	1.592
RMSEA	<0.08	0.041
IFI	>0.9	0.959
NFI	>0.9	0.897
TLI	>0.9	0.956
CFI	>0.9	0.959

Note:χ²/d f= Chi-squared degrees of freedom ratio; RMSEA = Root-mean-square error of approximation; IFI = Incremental fit index; NFI = Normed fit index; TLI = Tacker-Lewis index; CFI = Comparative fit index.

4.2 Hypotheses test results of direct paths

Figure 3 demonstrates the final estimation results of the structural model together with the standardized path coefficients and their respective significance levels. On this basis, the standardized path coefficients (β), critical ratio (C.R.) values, p-values, and hypothesis testing results of the model are presented more visually in Table 6. The path coefficients (β) reflect the relationship between the variables and their respective degrees of influence, while the C.R. determines the significance of regression coefficients. The C.R. value equal to or greater than 1.96 indicates significance at a p-value less than 0.05 [84].

Fig. 3

The structural model and impact paths.

Table 6

Summary table of path coefficients for fitted models

Direct path relationships hypothesized in the model	Coefficient (β)	S.E.	C.R.	P	Result
H1: SSL ⟶ ISSV	— —	— —	— —	— —
H1a: SCoa ⟶ ISSV	.033	.056	.676	.499	✗
H1b: SCon ⟶ ISSV	–.102	.056	–1.890	.059	✗
H1c: SCar ⟶ ISSV	–.161	.064	–2.440	.015	✓
H2: SSL ⟶ IRSV	— —	— —	— —	— —
H2a: SCoa ⟶ IRSV	–.159	.088	–2.556	.011	✓
H2b: SCon ⟶ IRSV	–.064	.087	–.948	.343	✗
H2c: SCar ⟶ IRSV	.135	.099	1.641	.101	✗
H3: SSL ⟶ SM	— —	— —	— —	— —	✓
H3a: Scoa ⟶ SM	.167	.058	3.375	^***	✓
H3b: SCon ⟶ SM	.232	.053	4.620	^***	✓
H3c: SCar ⟶ SM	.581	.052	11.172	^***	✓
H4: SM ⟶ SV	— —	— —	— —	— —
H4a: SM ⟶ ISSV	–.469	.080	–5.746	^***	✓
H4b: SM ⟶ IRSV	–.175	.116	– 1.822	.068	✗
H6: SSL ⟶ GSN	— —	— —	— —	— —
H6a: SCoa ⟶ GSN	.079	.065	1.453	.146	✗
H6b: SCon ⟶ GSN	.383	.063	6.630	^***	✓
H6c: SCar ⟶ GSN	.360	.054	6.729	^***	✓
H7: GSN ⟶ SV	— —	— —	— —	— —
H7a: GSN ⟶ ISSV	–.230	.054	–4.084	^***	✓
H7b: GSN ⟶ IRSV	–.046	.083	–.662	.508	✗
H9: SSL ⟶ SE	— —	— —	— —	— —
H9a: SCoa ⟶ SE	.199	.102	2.992	.003	✓
H9b: SCon ⟶ SE	–.015	.091	–.230	.818	✗
H9c: SCar ⟶ SE	.008	.081	.123	.902	✗
H10: SE ⟶ SV	— —	— —	— —	— —
H10a: SE ⟶ ISSV	–.005	.030	–.135	.893	✗
H10b: SE ⟶ IRSV	–.428	.050	–7.906	^***	✓

Note: ^*represents p < 0.05, ^**represents p < 0.01, ^***represents p < 0.001; Coefficient (β) = standardized regression weight; CR= critical ratio (> [1.96]); = hypothesis is partially supported; ✓= hypothesis is supported; û = hypothesis is rejected.

As seen in Fig. 3 and Table 6, among the three dimensions of SSL, only SCar exhibits a significant effect on ISSV (β= –0.159, p < 0.05). Therefore, only H1C holds for hypotheses H1a-c. Furthermore, only SCoa has an effect on IRSV (β= –0.161, p < 0.05). Therefore, only H2a holds true for hypotheses H2a-c. SCoa, SCon, and SCar all positively affect SM (β= 0.167, 0.232, 0.581, p < 0.001), confirming the validity of H3a-c. However, SM only influences ISSV (β= –0.469, p < 0.001), thus only supporting H4a in H4a-b. GSN is positively influenced by SCon and SCar (β= 0.383, 0.360, p < 0.001), thus validating H6b and H6c in H6a-c. However, GSN only influences ISSV (β= –0.230, p < 0.001), hence, only H7a is upheld within H7a-b. Finally, SE is only positively affected by SCoa (β= 0.199, p < 0.01). Thus, only H9a is supported within H9a-c. However, SE only influences IRSV (β= –0.428, p < 0.001), confirming only H10b in H10a-b.

4.3 Hypotheses test results of mediation paths

To understand the mediating role of workers’ safety motivation, group safety norms, and self-efficacy, the bootstrap method was used [86]. The mediating effect was tested by 2000 iterations, calculating 95% confidence intervals.

We provide a more intuitive list in Table 7, detailing the mediation effect values, 95% confidence intervals, and hypothesis testing results for each mediation pathway. For example, considering the SCoa⟶SM⟶ISSV path, the mediation effect value is –0.089. The upper and lower 95% confidence intervals exclude 0, and the p-value is less than the significant level of 0.05, indicating a mediation effect in the path. Similarly, the standardized indirect effects of SCon and SCar on ISSV through SM are (–0.113, p < 0.01, 95% CI [–0.207, –0.048]) and (–0.263, p < 0.01, 95% CI [–0.418, –0.146]), respectively. Therefore, hypotheses H5a, b, and c are all hold. Conversely, for the SCoa⟶SM⟶IRSV path, the mediation effect value is – 0.041. The upper and lower 95% confidence intervals contain 0, and the p-value is greater than the significance level of 0.05, indicating the absence of a mediation effect in this path. Similarly, the standardized indirect effects of SCon and SCar on IRSV through SM are not statistically significant (95% confidence interval including 0). Therefore, hypotheses H5d, e, and f are not hold. Hypothesis H5 is partially supported. In addition, the standardized indirect effects of SCon and SCar on ISSV through GSN are (–0.091, p < 0.01, 95% CI [–0.177, –0.036]) and (– 0.08, p < 0.01, 95% CI [–0.164, –0.028]), respectively. Thus, hypotheses H8b, and c are all hold. The standardized indirect effect of SCoa on ISSV through GSN and the standardized indirect effects of SCoa, SCon, and SCar on IRSV through GSN are not statistically significant (95% confidence interval including 0). Therefore, hypotheses H8a, d, e, and f are not hold. Hypothesis H8 is partially supported. Furthermore, the standardized indirect effect of SCoa on IRSV through SE is (– 0.121, p < 0.05, 95% CI [–0.226, –0.024]). Therefore, hypothesis H11d holds. However, the standardized indirect effects of SCoa, SCon and SCar on ISSV through SE and the standardized indirect effects of SCon and SCar on IRSV through SE are not statistically significant (95% confidence interval including 0). Therefore, hypotheses H11a, b, c, e, and f are not hold. Hypothesis H11 is partially supported.

Table 7
Summary of estimated indirect effects based on the bootstrap method

Mediation path relationships hypothesized in the model Coefficient 95% CI P Result

Lower Upper

H5: SSL ⟶ SM ⟶ SV — — — — — — — —

H5a: SCoa ⟶ SM ⟶ ISSV –.089 –.209 –.025 .005 ✓

H5b: SCon ⟶ SM ⟶ ISSV –.113 –.207 –.048 .001 ✓

H5c: SCar ⟶ SM ⟶ ISSV –.263 –.418 –.146 .001 ✓

H5d: SCoa ⟶ SM ⟶ IRSV –.041 –.132 .002 .042 ✗

H5e: SCon ⟶ SM ⟶ IRSV –.052 –.136 .002 .056 ✗

H5f: SCar ⟶ SM ⟶ IRSV –.122 –.310 .009 .074 ✗

H8: SSL ⟶ GSN ⟶ SV — — — — — — — —

H8a: SCoa ⟶ GSN ⟶ ISSV –.021 –.077 .011 .179 ✗

H8b: SCon ⟶ GSN ⟶ ISSV –.091 –.177 –.036 .001 ✓

H8c: SCar ⟶ GSN ⟶ ISSV –.080 –.164 –.028 .002 ✓

H8d: SCoa ⟶ GSN ⟶ IRSV –.005 –.051 –.008 .363 ✗

H8e: SCon ⟶ GSN ⟶ IRSV –.023 –.096 –.050 .474 ✗

H8f: SCar ⟶ GSN ⟶ IRSV –.020 –.091 .043 .452 ✗

H11: SSL ⟶ SE ⟶ SV — — — — — — — —

H11a: SCoa ⟶ SE ⟶ ISSV –.001 –.021 .024 .879 ✗

H11b: SCon ⟶ SE ⟶ ISSV .000 –.006 .008 .845 ✗

H11c: SCar ⟶ SE ⟶ ISSV .000 –.008 .006 .947 ✗

H11d: SCoa ⟶ SE ⟶ IRSV –.121 –.226 –.024 .013 ✓

H11e: SCon ⟶ SE ⟶ IRSV .008 –.065 .089 .800 ✗

H11f: SCar ⟶ SE ⟶ IRSV –.004 –.083 .072 .869 ✗

Mediation path relationships hypothesized in the model	Coefficient	95% CI	P	Result
H5: SSL ⟶ SM ⟶ SV	— —	— —	— —	— —
H5a: SCoa ⟶ SM ⟶ ISSV	–.089	–.209	–.025	.005	✓
H5b: SCon ⟶ SM ⟶ ISSV	–.113	–.207	–.048	.001	✓
H5c: SCar ⟶ SM ⟶ ISSV	–.263	–.418	–.146	.001	✓
H5d: SCoa ⟶ SM ⟶ IRSV	–.041	–.132	.002	.042	✗
H5e: SCon ⟶ SM ⟶ IRSV	–.052	–.136	.002	.056	✗
H5f: SCar ⟶ SM ⟶ IRSV	–.122	–.310	.009	.074	✗
H8: SSL ⟶ GSN ⟶ SV	— —	— —	— —	— —
H8a: SCoa ⟶ GSN ⟶ ISSV	–.021	–.077	.011	.179	✗
H8b: SCon ⟶ GSN ⟶ ISSV	–.091	–.177	–.036	.001	✓
H8c: SCar ⟶ GSN ⟶ ISSV	–.080	–.164	–.028	.002	✓
H8d: SCoa ⟶ GSN ⟶ IRSV	–.005	–.051	–.008	.363	✗
H8e: SCon ⟶ GSN ⟶ IRSV	–.023	–.096	–.050	.474	✗
H8f: SCar ⟶ GSN ⟶ IRSV	–.020	–.091	.043	.452	✗
H11: SSL ⟶ SE ⟶ SV	— —	— —	— —	— —
H11a: SCoa ⟶ SE ⟶ ISSV	–.001	–.021	.024	.879	✗
H11b: SCon ⟶ SE ⟶ ISSV	.000	–.006	.008	.845	✗
H11c: SCar ⟶ SE ⟶ ISSV	.000	–.008	.006	.947	✗
H11d: SCoa ⟶ SE ⟶ IRSV	–.121	–.226	–.024	.013	✓
H11e: SCon ⟶ SE ⟶ IRSV	.008	–.065	.089	.800	✗
H11f: SCar ⟶ SE ⟶ IRSV	–.004	–.083	.072	.869	✗

Note: = hypothesis is partially supported; ✓= hypothesis is supported; ✗= hypothesis is rejected.

5 Discussion

This research has constructed a theoretical model that explains the intrinsic relationship between frontline supervisors’ safety leadership and workers’ safety violations. Compared to previous studies [7], this study aims to gain a more detailed understanding of the intrinsic association between SSL and worker safety violations. By integrating the self-determination theory, social cognitive theory, and social information processing theory, we argue that, at the individual level, SSL enhances workers’ safety motivation (SM) and sense of self-efficacy (SE), thereby influencing workers’ safety violations. At the organizational level, SSL uses group safety norms (GSN) as a bridge to influence construction workers’ safety violations. Meanwhile, a theoretical model was constructed and validated by subdividing SSL into three dimensions: safety coaching (SCoa), safety controlling (SCon), and safety caring (SCar). Similarly, worker safety violations are subdivided into individuals’ situational safety violations (ISSV) and individuals’ routine safety violations(IRSV).

5.1 The direct relationship between SSL and safety violations

The tests conducted on hypotheses 1 and 2 show that only two paths, SCar⟶ISSV (H1c) and SCoa⟶IRSV (H2a), hold. Few studies have been able to account for this result. Therefore, we propose a possible explanation for the direct impact of safety caring on ISSV. As mentioned in Section 2.1., ISSV is caused by situational constraints during construction work (such as safety equipment not being easily accessible, etc.). We argue that resource support and emotional care from supervisors directly benefit workers. When supervisors provide care and support, workers’ safety needs are met and safeguarded, thereby removing situational constraints that impede their safety compliance [22]. As a result, their situational violations are diminished. In response to the direct influence of safety coaching on IRSV, we argue that within construction crews, workers tend to look to the supervisor as an opinion leader and seek their guidance to determine operational procedures [24]. Therefore, individual behavior is influenced by supervisors’ demonstration of safety behavior. It can be inferred that some of the individual’s undesirable behavior (e.g., failure to comply with safety procedures) also mostly comes from the imitation of supervisors’ erroneous safety coaching. Meanwhile, compared to other roles, the influence of supervisors is often more effective [43], which leaves a deeper impression on workers and leads to their daily and ongoing safety violations.

5.2 The mediating mechanisms between SSL and safety violations

5.2.1 The mediating effect of SM

The analysis of hypotheses H3, 4, and 5 found that all three dimensions of supervisor safety leadership positively affect workers’ safety motivation (SM). However, SM demonstrates a significant correlation only with ISSV, and it mediates the relationship between SCoa, SCon, SCar, and ISSV. However, we found notable differences in the effects of the three sub-dimensions of supervisor safety leadership on SM. Specifically, SCar exhibits a significantly stronger influence on SM compared to SCoa and SCon, with a path coefficient of 0.581 in contrast to 0.167 and 0.232 for the other two. Thus, there is a stronger association between supervisors’ safety caring and ISSV caused by situational constraints and their relationship is mainly mediated by the safety motivation of workers. That is, when supervisors care about their workers, provide safety resources, and seek their well-being, on the one hand, workers’ safety needs are more comprehensively guaranteed, so that workers do not fall into the embarrassment of passive violations due to lack of access to personal protective equipment (PPE) [39]. On the other hand, this supportive work environment also effectively motivates workers to engage in safety practices (e.g., helping coworkers obtain PPE in a more timely manner; giving their own and their coworkers’ safety needs timely feedback to their superiors, etc.) [39, 87]. As a result, workers’ situational safety violations are reduced. This finding not only supports the argument made by Chmiel et al. [35] that worker participation in non-coercive safety activities will reduce situational safety violations but also further elucidates that the distal antecedent of this mechanism is supervisors’ safetycaring.

5.2.2 The mediating effect of GSN

The analysis of hypotheses H6, 7, and 8 showed that two dimensions of supervisor safety leadership (SCon and SCar) positively influenced workers’ group safety norm (GSN) and that GSN mediated the relationship between SCon, SCar, and ISSV. This suggests that supervisors can influence worker safety behavior by establishing positive group norms. Although previous studies as well as our validation results suggest that care and support from supervisors reflect a strong supervisor-worker relationship and contribute to the creation of a favorable group climate [7]; our findings show that workers’ GSN is primarily influenced by supervisors’ safety controlling. This indicates that supervisors are more likely to play a more “assertive” role in creating an environment that places a high value on safety compliance [29]. Clearly, one way to achieve this is by exercising control over team members by fulfilling their roles and dictating what is expected of them (i.e., prohibitive norms), which seems to be more direct and effective [65]. In a group with high safety standards, workers will supervise each other and remind each other of precautions. Therefore, this effectively reduces violations caused by situational restrictions (e.g., forgetting to wear a safety helmet, etc.) [7]. Furthermore, the significant correlation between GSN and ISSV further supports the view that ISSV is closely related to workers’ psychosocial processes to a greater extent [7, 35].

5.2.3 The mediating effect of SE

The analysis of hypotheses H9, 10, and 11 showed that only SCoa positively influenced workers’ self-efficacy (SE), and SE mediated the relationship between SCoa and IRSV. SE reflects workers’ estimation and judgment regarding their ability to accomplish safety-related goals and tasks [67]. While supervisors’ safety coaching serves to exemplify workers’ behavior and mobilized workers’ intelligence, workers also benefit from substantial technical support [64]. This support fosters a greater sense of control over the safety situation at the site so as to enhance confidence levels. Under the effective coaching of supervisors, workers’ cognitive levels undergo a significant enhancement. Consequently, they carry out their daily tasks in a more standardized and quality-assured manner, rather than omitting some necessary safety steps in order to expedite tasks quickly or to save time and physical exertion. In addition, SE is only associated with IRSV, which further supports the argument made by Hansez and Chmiel [36] that routine safety violations involve a cognitive-energetic mechanism. High levels of SE among workers imply superior cognitive abilities, whereas high levels of SE derive from high levels of safety coaching from supervisors. Therefore, compared to previous studies [36, 88], our research findings may suggest a more complete pathway mechanism, namely: observational learning ⟶ ability enhancement ⟶ individual behavioral decision-making.

5.3 Theoretical contributions

The findings of this study yield significant theoretical contributions to the study of leadership behavior and occupational health and safety in the construction industry. While Current research on construction safety violations still focuses on organizational or individual factors, there have been relatively few studies on SSL. To the best of our knowledge, this is the first study to explore the multiple mediating mechanisms of the relationship between SSL and individuals’ safety violations from the perspective of construction workers. Therefore, this study offers some meaningful empirical insights into construction safety management. For example, three dimensions of SSL and three mediating variables were established to relate to different organizational and individual levels. This multidimensional analysis has addressed a limitation in previous safety leadership research, which tends to relate to organizational-level mediating variables. On this basis, the direct and indirect effects of three leadership characteristics on two types of safety violations (routine and situational) have been clarified. The results of the study highlight safety caring from supervisors can better stimulate the willingness of workers to actively engage in safety related actions, thereby reducing the probability of situational safety violations. In addition, supervisors’ safety controlling can be more effective in creating a good group safety climate. In such an organizational environment, workers are constantly subjected to good conduct and professional safety terms, which make them less susceptible to breaches of safety rules. Furthermore, supervisors’ safety coaching enhances workers’ competence and confidence, thereby promoting sustained adherence to safety protocols. As a result, this evidence-based research can inform the development of targeted interventions designed to effectively reduce both types of workers’ safety violations, namely, ISSV and IRSV. Therefore, this study clarifies the complex relationships between supervisor behavior and safety outcomes, particularly from the perspective of construction workers and their perceptions of supervisors’ actions and leadership.

5.4 Practical implications

This study offers practical implications for construction safety management. Firstly, top management in the construction industry should prioritize education and training for frontline supervisors to enhance their skills, safety competence, and correct safety behavior [89]. In this way, these principles penetrate the organization from the top down, and the workers learn through emulation coupled with the supervisors’ correct safety coaching; workers can significantly improve their safety competence and confidence to ensure diligent performance of safety duties. Secondly, supervisors should create a positive safety climate at the workplace by communicating the principle of “safety first” to workers. This involves a pattern of high expectations, high standards of supervision, and rewards and penalties. Through the establishment of “prohibitive norms” within the organization, worker groups can demonstrate high standards of safety behavior, and workers will remind and help each other to adhere to these regulations. Moreover, supervisors should increase their caring for workers’ living conditions, take flexible measures to meet sensible and reasonable work requirements within the organizational framework, offer increased encouragement to employees and humbly accept workers’ safety-related suggestions. The integration of humanistic care and respect for individuality within institutional constraints greatly improves workers’ satisfaction and recognition of their supervisors [90]. This approach also better shapes workers’ safety values and creates stronger motivation among employees to prioritize safety in their work.

6 Conclusion

Using a theory-driven model, this study uncovered the impact of different dimensions of SSL (namely, safety coaching, safety controlling, and safety caring) on two types of safety violations (namely, routine and situational) from the perspective of construction workers. Safety coaching directly predicted workers’ routine safety violations, while safety caring directly predicted situational safety violations. More importantly, different types of SSL can affect different types of safety violations through different mediators. Specifically, workers’ safety motivation was more likely to be influenced by supervisors’ safety caring, which reduced their situational safety violations; the establishment of group safety norms among workers depended primarily on supervisors’ safety controlling, which further reduced their situational safety violations; and supervisors’ safety coaching influenced routine violations by acting primarily on workers’ self-efficacy. Safety motivation, group safety norms, and self-efficacy play important roles in the different pathways between SSL and safety violations and serve as significant predictors of safety violations. These findings suggest that interventions from a supervisor’s perspective can effectively guide workers to develop good safety intentions, boost their confidence in working safely and create a favorable group safety climate. Consequently, such interventions contribute to the establishment of a more stable and secure construction work environment.

Some limitations should be acknowledged in this study. Firstly, the generalizability of the results may be limited as the survey was confined to certain regions of China with a single industry. Future research could explore safety violations in other industries or cultures to gain a deeper understanding of safety violations and effective safety leadership through a more diverse sample. Secondly, the data collection of the sample was done using the cross-sectional method, which may be prone to social desirability biases. Factors such as respondents’ personality traits, cognitive processes, job-related sentiments, different groups, and social influences from others may affect the measurement results. Future studies could conduct more in-depth interviews with program staff to reveal their differential perceptions regarding their approach to the problem and weigh them in the design of the questionnaire.

Ethical statement

This research was approved by Research Ethics Committee, Jiangsu University. This is an original work that has not been published or submitted for publication elsewhere. The authors understand that submission of the paper means the copyright transfer from the authors to the publisher of WORK: A Journal of Prevention, Assessment & Rehabilitation.

Informed consent

Informed consent was obtained from all participants. The manuscript does not contain any identifiable individual data in any form.

Conflict of interest

The authors declare that they have no conflict of interest.

Footnotes

Acknowledgments

The authors would like to thank the anonymous reviewers for their invaluable and constructive comments.

Funding

This work was funded by the National Natural Science Foundation of China (Nos. 72071096, 71971100, 72101055); sponsored by Qing Lan Project of Jiangsu Province.

Appendix. Items of Scales Developed for the Study

References

Wang

, Zou

PXW

, Li

. Critical factors and paths influencing construction workers’ safety risk tolerances. Accid Anal Prev. 2016;93:267–79.

Meng

, Liu

, Li

, Hu

. Influencing Factors, Mechanism and Prevention of Construction Workers’ Unsafe Behaviors: A Systematic Literature Review. Int J Environ Res Public Health. 2021;18(5):22.

Seo

, Han

, Lee

, Kim

. Computer vision techniques for construction safety and health monitoring. Adv Eng Inform. 2015;29(2):239–51.

Zhang

, Shi

, Yang

. A critical review of vision-based occupational health and safety monitoring of construction site workers. Saf Sci. 2020;126:18.

Shen

, Ju

, Koh

, Rowlinson

, Bridge

. The Impact of Transformational Leadership on Safety Climate and Individual Safety Behavior on Construction Sites. Int J Environ Res Public Health. 2017;14(1).

Guo

BHW

, Yiu

, Gonzalez

. Predicting safety behavior in the construction industry: Development and test of an integrative model. Saf Sci. 2016;84:1–11.

, Cong

, Liang

. The impact of supervisor-worker relationship on workers’ safety violations: A modified theory of planned behaviour. J Civ Eng Manag. 2019;25(7):631–45.

Suraji

, Duff

, Peckitt

. Development of causal model of construction accident causation. J Constr Eng Manage-ASCE. 2001;127(4):337–44.

Hinze

, Rinker

. Construction safety. Saf Sci. 2008;46(4):565–65.

10.

Xia

, Zou

PXW

, Liu

, Wang

, Zhu

. A hybrid BN-HFACS model for predicting safety performance in construction projects. Saf Sci. 2018;101:332–43.

11.

Garrett

, Teizer

. Human factors analysis classification system relating to human error awareness taxonomy in construction safety. J Constr Eng Manage. 2009;135(8):754–63.

12.

Wong

, Wang

, Law

, Lo

. Association of root causes in fatal fall-from-height construction accidents in Hong Kong. J Constr Eng Manage. 2016;142(7).

13.

Hon

CKH

, Chan

APC

, Wong

FKW

. An analysis for the causes of accidents of repair, maintenance, alteration and addition works in Hong Kong. Saf Sci. 2010;48(7):894–901.

14.

Choudhry

, Fang

, Mohamed

. The nature of safety culture: A survey of the state-of-the-art. Saf Sci. 2007;45(10):993–1012.

15.

Barling

, Loughlin

, Kelloway

. Development and test of a model linking safety-specific transformational leadership and occupational safety. Journal of Applied Psychology. 2002;87(3):488.

16.

T-C

. The validity and reliability of safety leadership scale in universities of Taiwan. International Journal of Technology Engineering Education. 2005;2(1).

17.

, Fang

, Li

. Roles of owners’ leadership in construction safety: The case of high-speed railway construction projects in China. Int J Proj Manag. 2015;33(8):1665–79.

18.

, Li

, Fang

. Leadership improvement and its impact on workplace safety in construction projects: A conceptual model and action research. Int J Proj Manag. 2017;35(8):1495–511.

19.

, Wang

, Zou

PXW

, Fang

. How safety leadership works among owners, contractors and subcontractors in construction projects. Int J Proj Manag. 2016;34(5):789–805.

20.

Flin

, Yule

. Leadership for safety: industrial experience. Quality & Safety in Health Care. 2004;13:45–51.

21.

Fang

, Wu

. Impact of the Supervisor on Worker Safety Behavior in Construction Projects. J Manage Eng. 2015;31(6):12.

22.

Puah

, Ong

, Chong

. The effects of perceived organizational support, perceived supervisor support and perceived co-worker support on safety and health compliance. Int J Occup Saf Ergon. 2016;22(3):333–39.

23.

Sasser

Jr , Leonard

. Let first-level supervisors do their job. Harvard Business Review. 1980;58(2):113–21.

24.

Lingard

, Cooke

, Blismas

. Do perceptions of supervisors’ safety responses mediate the relationship between perceptions of the organizational safety climate and incident rates in the construction supply chain? J Constr Eng Manage. 2012;138(2):234–41.

25.

Zhang

, Li

, Fang

, Wu

. Supervisor-focused behavior-based safety method for the construction industry: case study in Hong Kong. J Constr Eng Manage. 2017;143(7).

26.

Lee

C-C

, Yeh

W-C

, Yu

, Tsai

T-H

. The impacts of supervisor support, role perception, and emotional exhaustion on the turnover intentions of real estate brokers. Scandinavian Journal of Management. 2022;38(4).

27.

, Yue

, Yang

, Li

, Xiang

, Fu

, Cui

. Understanding the sociocognitive process of construction workers’ unsafe behaviors: An agent-based modeling approach. Int J Environ Res Public Health. 2020;17(5):33.

28.

Kapp

. The influence of supervisor leadership practices and perceived group safety climate on employee safety performance. Saf Sci. 2012;50(4):1119–24.

29.

Clarke

. Safety leadership: A meta-analytic review of transformational and transactional leadership styles as antecedents of safety behaviours. J Occup Organ Psychol. 2013;86(1):22–49.

30.

Draghici

, Dursun

, Basol

, Boatca

, Gaureanu

. The mediating role of safety climate in the relationship between transformational safety leadership and safe behavior-the case of two companies in Turkey and Romania. Sustainability. 2022;14(14):18.

31.

Wang

, Mao

, Zhao

, Wang

, Hu

. The cross-level effect of team safety-specific transformational leadership on workplace safety behavior: The serial mediating role of team safety climate and team safety motivation. J Saf Res. 2023;87:285–96.

32.

, Wang

, Ruan

, Zheng

. Divergent effects of transformational leadership on safety compliance: A dual-path moderated mediation model. Plos One. 2022;17(1):17.

33.

T-C

. Safety leadership in the teaching laboratories of electrical and electronic engineering departments at Taiwanese Universities. J Saf Res. 2008;39(6):599–607.

34.

T-C

, Chen

C-H

, Li

C-C

. A correlation among safety leadership, safety climate and safety performance. Journal of Loss Prevention in the Process Industries. 2008;21(3):307–18.

35.

Chmiel

, Laurent

, Hansez

. Employee perspectives on safety citizenship behaviors and safety violations. Saf Sci. 2017;93:96–107.

36.

Hansez

, Chmiel

. Safety behavior: job demands, job resources, and perceived management commitment to safety. J Occup Health Psychol. 2010;15(3):267–78.

37.

Laurent

, Chmiel

, Hansez

. Jobs and safety: A social exchange perspective in explaining safety citizenship behaviors and safety violations. Saf Sci. 2018;110:291–9.

38.

Smith

, Eldridge

, DeJoy

. Safety-specific transformational and passive leadership influences On firefighter safety climate perceptions and safety behavior outcomes. Saf Sci. 2016;86:92–7.

39.

Liang

, Lin

, Zhang

, Su

. The impact of coworkers’ safety violations on an individual worker: A social contagion effect within the construction crew. Int J Environ Res Public Health. 2018;15(4):21.

40.

Liang

, Zhang

. Impact of supervisors’ safety violations on an individual worker within a construction crew. Saf Sci. 2019;120:679–91.

41.

Lawton

. Not working to rule: understanding procedural violations at work. Saf Sci. 1998;28(2):77–95.

42.

Man

, Chan

AHS

, Wong

. Risk-taking behaviors of Hong Kong construction workers – A thematic study. Saf Sci. 2017;98:25–36.

43.

, Qian

, Zhang

. Impact of supervisor leadership on construction worker safety behavior in China: the moderating role of social capital. Eng Constr Archit Manag. 2022:26.

44.

Dickerson

, Koch

, Adams

, Goodfriend

, Donnelly

. Safety coaches in radiology: decreasing human error and minimizing patient harm. Pediatr Radiol. 2010;40(9):1545–51.

45.

, Yu

. The relationships among perceived patients’ safety culture, intention to report errors, and leader coaching behavior of nurses in Korea: A pilot study. J Patient Saf. 2017;13(3):175–83.

46.

Petitta

, Probst

, Barbaranelli

, Ghezzi

. Disentangling the roles of safety climate and safety culture: Multi-level effects on the relationship between supervisor enforcement and safety compliance. Accid Anal Prev. 2017;99:77–89.

47.

Liu

, Yang

, Mei

. The effect of perceived organizational support for safety and organizational commitment on employee safety behavior: a meta-analysis. Int J Occup Saf Ergon. 2021;27(4):1154–65.

48.

Hedlund

, Gummesson

, Rydell

, Andersson

. Safety motivation at work: Evaluation of changes from six interventions. Saf Sci. 2016;82:155–63.

49.

Jiang

, Probst

. Transformational and passive leadership as cross-level moderators of the relationships between safety knowledge, safety motivation, and safety participation. J Saf Res. 2016;57:27–32.

50.

Quansah

, Zhu

, Obeng

. Assessment of the effects of supervisor behaviour, safety motivation and perceived job insecurity on underground miner’s safety citizenship behaviour. Chin Manag Stud. 2022;16(2):356–81.

51.

Kovjanic

, Schuh

, Jonas

, Van Quaquebeke

, Van Dick

. How do transformational leaders foster positive employee outcomes? A self-determination-based analysis of employees’ needs as mediating links. Journal of Organizational Behavior. 2012;33(8):1031–52.

52.

Probst

, Brubaker

. The effects of job insecurity on employee safety outcomes: cross-sectional and longitudinal explorations. J Occup Health Psychol. 2001;6(2):139.

53.

Ansori

, Widyanti

, Yassierli . The Influence of Safety Climate, Motivation, and Knowledge on Worker Compliance and Participation: An Empirical Study of Indonesian SMEs. Ing Invest. 2021;41(3):9.

54.

Maneechaeye

, Potipiroon

. The impact of fleet-level and organization-level safety climates on safety behavior among Thai civilian pilots: The role of safety motivation. Saf Sci. 2022; 147:9.

55.

Subramaniam

, Johari

, Mashi

, Mohamad

. The influence of safety leadership on nurses? Safety behavior: The mediating role of safety knowledge and motivation. J Saf Res. 2023;84:117–28.

56.

Van den Broeck

, Ferris

, Chang

, Rosen

. A review of self-determination theory’s basic psychological needs at work. Journal of Management. 2016;42(5):1195–229.

57.

Rivis

, Sheeran

. Descriptive norms as an additional predictor in the theory of planned behaviour: A meta-analysis. Current Psychology. 2003;22(3):218–33.

58.

Choi

, Lee

Modeling the effect of a socio-psychological process on construction workers’ safety behavior. in ASCE International Workshop on Computing in Civil Engineering (IWCCE). 2017. Univ Washington, Seattle, WA. .

59.

McFadden

, Stock

, Gowen

. Leadership, safety climate, and continuous quality improvement: Impact on process quality and patient safety. Health Care Management Review. 2015;40(1):24–34.

60.

, McCabe

, Jia

. Effect of leader-member exchange on construction worker safety behavior: Safety climate and psychological capital as the mediators. Saf Sci. 2021;142.

61.

Choi

, Ahn

, Lee

. Construction Workers’ Group Norms and Personal Standards Regarding Safety Behavior: Social Identity Theory Perspective. J Manage Eng. 2017;33(4):11.

62.

Roettger

, Maier

, Krex-Brinkmann

, Kowalski

, Krick

, Felfe

, Stein

. Social cognitive aspects of the participation in workplace health promotion as revealed by the theory of planned behavior. Preventive Medicine. 2017;105:104–8.

63.

Chen

C-F

, Chen

S-C

. Measuring the effects of Safety Management System practices, morality leadership and self-efficacy on pilots’ safety behaviors: Safety motivation as a mediator. Saf Sci. 2014;62:376–85.

64.

Bono

, Ilies

. Charisma, positive emotions and mood contagion. Leadership Quarterly. 2006;17(4):317–34.

65.

Fugas

, Silva

, Meliá

. Another look at safety climate and safety behavior: Deepening the cognitive and social mediator mechanisms. Accid Anal Prev. 2012;45:468–77.

66.

Sun

, Xia

. Teacher-perceived distributed leadership, teacher self-efficacy and job satisfaction: A multilevel SEM approach using the TALIS data. International Journal of Educational Research. 2018;92:86–97.

67.

Yen

C-D

, Hsu

S-W

, Ye

K-D

, Shang

K-C

, Wu

C-H

. A safety compliance model for container terminal operations: the roles of the safety climate, non-compliance cost and safety self-efficacy. Maritime Policy & Management. 2024;51(1):118–32.

68.

Saleem

, Isha

ASN

, Yusop

, Awan

, Naji

GMA

. The role of psychological capital and work engagement in enhancing construction workers’ safety behavior. Front Public Health. 2022;10.

69.

Kim

, Jang

. Seafarers’ quality of life: Organizational culture, self-efficacy, and perceived fatigue. Int J Environ Res Public Health. 2018;15(10):11.

70.

Salancik

, Pfeffer

. A social information processing approach to job attitudes and task design. Administrative Science Quarterly. 1978:224–53.

71.

Zalesny

, Ford

. Extending the social information processing perspective: New links to attitudes, behaviors, and perceptions. Organizational Behavior and Human Decision Processes. 1990;47(2):205–46.

72.

Wadei

, Chen

, Frempong

, Appienti

. The mediation effect of ethical leadership and creative performance: A social information processing perspective. J Creat Behav. 2021;55(1):241–54.

73.

Boekhorst

. The role of authentic leadership in fostering workplace inclusion: A social information processing perspective. Hum Resour Manage. 2015;54(2):241–64.

74.

Nykanen

, Salmela-Aro

, Tolvanen

, Vuori

. Safety self-efficacy and internal locus of control as mediators of safety motivation – Randomized controlled trial (RCT) study. Saf Sci. 2019;117:330–8.

75.

Luthans

, Avolio

, Avey

, Norman

. Positive psychological capital: Measurement and relationship with performance and satisfaction. Personnel Psychology. 2007;60(3):541–72.

76.

Hae-Young

. Statistical notes for clinical researchers: assessing normal distribution (2) using skewness and kurtosis. Restorative dentistry endodontics. 2013;38(1):52–4.

77.

Pektas

. Determining the essential parameters of bed load and suspended sediment load. International Journal of Global Warming. 2015;8(3):335–59.

78.

, Wu

, Wang

, Hu

. Relationship between Social Capital, Safety Competency, and Safety Behaviors of Construction Workers. J Constr Eng Manage. 2020;146(6):10.

79.

Hair

. Multivariate data analysis. 2009.

80.

Black

, Babin

, Multivariate data analysis: Its approach, evolution, and impact, in The Great Facilitator: Reflections on the Contributions of Joseph F Hair, Jr to Marketing and Business Research,

Black and

B.J.

Babin , Editors. 2019, Springer. p. 121–30.

81.

Fornell

, Larcker

. Structural equation models with unobservable variables and measurement error: algebra and statistics. Journal of Marketing Research. 1981;18(1):39–50.

82.

Bagozzi

. Evaluating structural equation models with unobservable variables and measurement error: A Comment. Journal of Marketing Research. 1981;18(3):375–81.

83.

Bagozzi

, Yi

. On the evaluation of structural equation models. Journal of the Academy of Marketing Science. 1988;14(1):33–46.

84.

Marsh

, Hau

. Assessing goodness of fit. Journal of Experimental Education. 1995;64(4):364–90.

85.

Xiong

, Skitmore

, Xia

. A critical review of structural equation modeling applications in construction research. Autom Constr. 2015;49:59–70.

86.

Efron

. Bootstrap methods: Another look at the jackknife. The Annals of Statistics. 1979;7(1):1–26.

87.

, Griffin

, Bertuleit

. Modelling antecedents of safety compliance: Incorporating theory from the technological acceptance model. Saf Sci. 2016;87:292–8.

88.

Duff

, Podolsky

, Biron

, Chan

CCA

. The interactive effect of team and manager absence on employee absence: A multilevel field study. J Occup Organ Psychol. 2015;88(1):61–79.

89.

Chih

Y-Y

, Kiazad

, Cheng

, Lajom

JAL

, Restubog

SLD

. Feeling positive and productive: Role of supervisor-worker relationship in predicting construction workers’ performance in the Philippines. J Constr Eng Manage. 2017;143(8).

90.

Fan

, Lo

CKY

, Ching

, Kan

. Occupational health and safety issues in operations management: A systematic and citation network analysis review. Int J Prod Econ. 2014;158:334–44.

Variable	Indicator	SFL	AVE	CR	α
Safety coaching	SCoa1	0.823	0.683	0.928	0.926
	SCoa2	0.858
	SCoa3	0.829
	SCoa4	0.742
	SCoa5	0.830
	SCoa6	0.871
Safety controlling	SCon1	0.789	0.629	0.922	0.922
	SCon2	0.797
	SCon3	0.806
	SCon4	0.807
	SCon5	0.844
	SCon6	0.733
	SCon7	0.770
Safety caring	SCar1	0.811	0.714	0.937	0.937
	SCar2	0.804
	SCar3	0.769
	SCar4	0.951
	SCar5	0.797
	SCar6	0.920
Safety motivation	SM1	0.856	0.695	0.872	0.872
	SM2	0.851
	SM3	0.793
Group safety norm	GSN1	0.822	0.700	0.921	0.920
	GSN2	0.873
	GSN3	0.811
	GSN4	0.822
	GSN5	0.855
Self-efficacy	SE1	0.852	0.749	0.947	0.947
	SE2	0.839
	SE3	0.912
	SE4	0.852
	SE5	0.851
	SE6	0.885
Individuals’ routine safety violations	IRSV1	0.840	0.728	0.914	0.913
	IRSV2	0.901
	IRSV3	0.815
	IRSV4	0.854
Individuals’ situational safety violations	ISSV1	0.783	0.652	0.918	0.915
	ISSV2	0.777
	ISSV3	0.848
	ISSV4	0.785
	ISSV5	0.821
	ISSV6	0.829

Multidimensional analysis of supervisors’ safety leadership on safety violations of construction workers: An empirical investigation

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSION:

Keywords

1 Introduction

2 Literature review

2.1 Safety violations

2.2 Supervisors’ safety leadership (SSL)

2.2.1 Definition of SSL

2.2.2 Dimensions of SSL

2.3 SSL and safety violations

2.4 Safety motivation as the mediating variable

2.5 Group safety norm as the mediating variable

2.6 Self-efficacy as the mediating variable

3.1 Participants and procedures

3.3.1 Descriptive analyses

4.1 Structural model fit

Table 5 Model goodness-of-fit index Fit Metrics Acceptable range Measurements χ2/df <3 1.592 RMSEA <0.08 0.041 IFI >0.9 0.959 NFI >0.9 0.897 TLI >0.9 0.956 CFI >0.9 0.959

5.1 The direct relationship between SSL and safety violations

5.2 The mediating mechanisms between SSL and safety violations

5.2.1 The mediating effect of SM

5.2.2 The mediating effect of GSN

5.2.3 The mediating effect of SE

5.3 Theoretical contributions

5.4 Practical implications

6 Conclusion

Ethical statement

Informed consent

Conflict of interest

Footnotes

Acknowledgments

Funding

Appendix. Items of Scales Developed for the Study

References

Table 5
Model goodness-of-fit index

Fit Metrics Acceptable range Measurements

χ²/df <3 1.592

RMSEA <0.08 0.041

IFI >0.9 0.959

NFI >0.9 0.897

TLI >0.9 0.956

CFI >0.9 0.959