Enhancing Sustainable Construction Safety: A Self-Determination Theory Approach to Worker Safety Behavior

Abstract

The construction industry has long been recognized as one of the world’s most hazardous sectors, with safety issues remaining a persistent challenge. To enhance sustainable safety management in this field, this study employs Self-Determination Theory (SDT) to explore the psychological mechanisms underlying construction workers’ safety behaviors. Through structural equation modeling using SPSS 27 and AMOS 28 software on 334 questionnaires, the research reveals that safety awareness and work pressure not only directly influence safety behaviors but also mediate through psychological factors. Notably, social identity significantly moderates the cognitive–behavioral pathway, while autonomous and control motivations moderate the psychological–behavioral relationship. This study breaks through the limitations of traditional safety management that focuses solely on external factors, constructing a multi-level theoretical model that encompasses cognitive, stress, psychological, motivational, and social factors. The research provides a theoretical foundation and practical pathway for construction enterprises to implement people-oriented safety management, contributing to the Sustainable Development Goals centered on the health and safety of employees.

1. Introduction

The construction industry has long been recognized as one of the most dangerous globally, with safety issues remaining a major pain point in this field [1]. Although data from China’s Ministry of Housing and Urban–Rural Development shows that construction safety accidents decreased in 2024, the safety situation in the industry remains severe [2]. Due to the industry’s unique attributes and the characteristics of its workforce, the accident rate in construction is significantly higher than in most other industries [3]. On construction sites, workers are both the main participants and direct victims of safety incidents, with over 70% of accidents attributed to unsafe work behaviors [4,5]. Therefore, workers’ safety behaviors play a crucial role in reducing accidents and safeguarding life and health [6]. Promoting safe production is not only a moral responsibility but also closely related to the achievement of Sustainable Development Goals (SDGs)—particularly promoting Sustainable Development Goalal 8 (SDG8, decent work and protecting workers’ safety) and Sustainable Development Goal 3 (SDG3, improving health and well-being). Despite continuous advancements in safety management technologies, construction sites remain high-risk areas [7]. This ongoing loss of life not only incurs huge economic costs and family traumas but also severely hinders the sustainable progress of society, economy, and environment. Therefore, how to effectively prevent and reduce accidents on construction sites has become an important topic of common concern in both academic circles and industry practices.

In the field of construction safety research, existing literature mainly explores two aspects: one part focuses on external intervention measures for safety management [8], such as regulations and policies [9], technical protection [10], and operation training [11], which have laid an important foundation for building a basic safety environment; meanwhile, another part gradually shifts to the exploration of the internal formation mechanism of behavior [12], especially after the introduction of self-determination theory, which provides a theoretical perspective for understanding how individual motivation affects safety behavior. This theory holds that such systems can stimulate the intrinsic motivation for safety behavior by meeting workers’ needs for autonomy and competence. Applying this theory to the study of construction workers’ safety behavior can reveal the intrinsic driving mechanism from a psychological perspective and provide new ideas and methods for improving safety management. However, although self-determination theory (SDT) has been applied in workplace safety research, there is a lack of comprehensive examination of the overall transformation mechanism of “cognition–stress–psychology–behavior”, especially ignoring the synergistic role of social identity and motivation types in behavior formation. This study, grounded in SDT, constructs a multi-level theoretical framework encompassing cognitive, stress, psychological, motivational, and social factors. The research not only validates the critical mediating role of safety psychology in the relationship between safety cognition, occupational stress, and safety behavior but also pioneers the integration of social identity and motivational types into a unified analytical system. It reveals the moderating effect of social identity in the cognitive-to-behavioral transformation process, while elucidating the differential impacts of various motivational types on the relationship between psychological states and behavioral outcomes.

The remainder of this paper is structured as follows. First, we conduct a literature review to develop our hypotheses, focusing on the key elements of our theoretical solution: safety cognition, work pressure, safety psychology, social identity, and autonomous/controlled motivation. Subsequently, we detail the methodology, including the sample, measures, and analytical procedures used to test our model. We then present the results of the hypothesis testing, mediation, and moderation analyses. Finally, we discuss the theoretical and practical implications of our findings, contextualize them within the broader literature, acknowledge the study’s limitations, and suggest directions for future research.

2. Literature Review and Hypotheses

2.1. Self-Determination Theory

Compared to alternative modeling approaches like neural networks, our choice of the SDT model is driven by research objectives. While neural networks excel at predicting complex nonlinear data patterns, they struggle to uncover the causal mechanisms underlying behaviors. In contrast, the SDT model provides a concise and interpretable theoretical framework that allows us to formulate specific hypotheses regarding the psychological needs and motivational processes behind safety behavior changes. This theory-driven research methodology is crucial for developing interventions, as it not only reveals phenomena but also clarifies their causes—for example, enhancing autonomy proves more effective than simply imposing external pressure. Therefore, the SDT model is better suited for explaining the intrinsic psychological mechanisms of safety behaviors, while neural networks are more appropriate for studies focused solely on achieving maximum accuracy in predicting safety events. SDT, proposed by Deci and Ryan in 1985, offers a systematic framework for understanding human motivation and personality development. The theory posits that individual behavior is influenced by both intrinsic and extrinsic motivations, with intrinsic motivation closely related to the satisfaction of basic psychological needs such as autonomy, competence, and relatedness, while extrinsic motivation mainly depends on external reward and punishment mechanisms [13,14,15,16]. While traditional frameworks like the Theory of Planned Behavior (TPB) primarily focus on conscious behavioral beliefs and external social pressures, the SDT provides a more nuanced explanation through the lens of basic psychological needs. The unique value of SDT lies in its ability to clarify how external safety regulations and norms are internalized as personal values. The model posits that when the work environment supports employees’ intrinsic psychological needs for autonomy, competence, and relatedness, external safety values are more likely to transform into fully internalized intrinsic motivation. This internalization process is the key mechanism that distinguishes SDT from traditional models.

From the perspective of SDT, fulfilling the psychological needs of construction workers in three dimensions—autonomy [17], competence [18], and relatedness [19]—can effectively enhance their safety behaviors. Meeting employees’ basic psychological needs not only significantly improves job satisfaction and intrinsic motivation but also serves as a key factor in explaining and predicting their behavioral patterns. Understanding these psychological drivers is crucial not only for immediate safety outcomes but also directly supports the sustainable development of the construction industry. Autonomy refers to workers’ sense of control and freedom over their actions. Involving employees in safety decision-making and granting them autonomy can significantly enhance their sense of responsibility and initiative in safety behaviors. This is essential for building a socially sustainable operational system, enabling employees to truly become active participants in their own well-being [17]. Competence involves employees’ awareness and confidence in their abilities. Systematic safety training and skill enhancement can effectively reduce errors and rework. This not only helps conserve resources and minimize waste, thereby directly promoting economic sustainability, but also avoids material damage and corrective work, thus advancing environmental sustainability [18]. Relatedness concerns employees’ sense of belonging and support within teams. A positive team atmosphere and management support can strengthen employees’ sense of belonging, thereby improving their safety behaviors [19]. Furthermore, this study aligns closely with the principles of sustainable human resource management. This philosophy positions employees’ long-term development and well-being as the organization’s core assets [20]. Sustainable HRM transcends traditional performance metrics by incorporating employees’ physical and mental health into its scope. In the construction safety sector, this approach advocates creating work environments that fulfill employees’ fundamental psychological needs—namely, autonomy, competence, and a sense of belonging. By investing in such psychologically supportive safety practices, companies can not only enhance immediate safety performance but also foster employees’ long-term mental health and sustainable employability, thereby achieving key social sustainability objectives.

2.2. Safety Cognition and Safety Behavior

Cognitive psychology indicates that human behavior is the product of cognition [21]. According to Neisser, “cognition” refers to the process by which all sensory inputs are transformed, simplified, processed, stored, retrieved, and utilized [22]. Combining the research on cognition and safety cognition, safety cognition is defined as the process of obtaining and processing information related to safety or danger and implementing decisions [23].

From the perspective of construction sites, safety cognition is the core process by which construction workers acquire relevant safety information on the site, understand potential risks, choose safe responses, and execute safe operations. Existing research has clearly pointed out that the unsafe behaviors of construction workers often stem directly from mistakes in the safety cognition process [24]. Whether it is the failure to accurately capture on-site danger information through the senses, the misunderstanding of the obtained safety information, or the deviation from safety logic in the response selection and execution stages, all can directly lead to unsafe operations [25]. Further, in the context of high-altitude work, workers’ accurate cognition of high-altitude dangers can directly reduce unsafe behaviors related to falls [26]. It is evident that safety cognition positively influences the safety behaviors of construction workers by directly affecting their risk perception and operation choices. Based on the above literature analysis, we propose the following hypothesis:

H1: 

Safety cognition has a positive impact on the safety behaviors of construction workers.

2.3. Safety Psychology and Safety Behavior

Safety psychology examines the comprehensive psychological state formed by individuals in specific work environments, based on their perceptions of organizational safety culture, managerial support, and their own roles. This state comprises three core dimensions: safety attitudes, safety self-efficacy, and safety-specific emotions. It directly influences individuals’ risk assessment and safety decision-making, serving as the core cognitive mechanism that drives construction workers to transition from passive compliance to proactive safety practices [27,28].

At the same time, cognitive psychology holds that the core of human psychology and behavior is cognitive issues. Human behavior is determined by the interaction between the behavioral environment and the self. Personal behavior is not an isolated and simple response to external stimuli, nor is it the mechanical sum of many reflex arcs. Instead, it is made through the integration of the psychophysical field [17]. Specifically, when workers develop a high level of identification with the project, they internalize safety rules as their own value standards. This cognitive-level internalization significantly enhances the self-awareness of safety behaviors [27]. Additionally, research indicates that self-efficacy in safety psychology directly influences behavioral decisions; that is, the stronger a worker’s perception of their own safety capabilities, the more likely they are to proactively avoid risks [29]. Moreover, cognitive-level safety motivation has been confirmed as the key path connecting psychological safety and safety behaviors—when workers perceive organizational support, a clear association of “safety behaviors are beneficial” forms in their cognition, enabling them to spontaneously comply with safety norms without external incentives [28]. This direct mechanism of cognition on behavior indicates that improving safety psychology is not merely through emotional regulation but through reshaping workers’ cognitive evaluations of safety behaviors. Based on the above literature analysis, we propose the following hypothesis:

H2: 

Safety psychology has a positive impact on the safety behaviors of construction workers.

2.4. Work Pressure and Safety Behavior

Work pressure is defined as the subjective perception of stress caused by high-intensity working conditions. It is manifested in three aspects: excessive workload, sense of time urgency and poor physical environment [30,31].

Furthermore, workplace stress frequently emerges as a focal point in occupational health discussions, particularly regarding its significant impact on safety compliance [32]. Research indicates that high-intensity work demands coupled with resource shortages exhibit a negative correlation with safety compliance. This implies that when employees experience excessive work pressure, they may fail to effectively adhere to safety protocols, thereby increasing accident risks [12]. Moreover, the cognitive depletion caused by stress leads employees to prioritize efficiency over safety, often resorting to shortcuts like ignoring safety regulations to expedite task completion [33]. The substantial adverse effects of work pressure on safety behaviors also pose threats to economic and environmental sustainability. Errors and accidents resulting from high-pressure environments may lead to increased costs and material waste, negatively impacting project efficiency and environmental performance. More critically, persistent work pressure reinforces the erroneous belief that “safety compromises efficiency,” creating a self-perpetuating cycle where employees repeatedly engage in high-risk behaviors in subsequent operations. Based on the above literature analysis, we propose the following hypothesis:

H3: 

Work pressure has a negative impact on the safety behavior of construction workers.

2.5. The Mediating Effect of Safety Psychology

Safety psychology not only refers to the psychological state when engaging in a certain job but also encompasses an individual’s awareness, knowledge, attitude, and behavior towards safety issues. These psychological activities may influence the probability of safety accidents occurring [34].

Specifically, when workers identify potential dangers through perception and understanding, this cognitive result triggers their safety motivation, safety attitude, and risk perception at the psychological level. These psychological responses determine whether they are willing to invest additional effort in taking protective measures [35]. In other words, if the psychological response is positive, cognition will be transformed into safety behavior; if the psychological response is negative, even if the cognition is in place, the behavior may be inhibited [36]. This mechanism has been verified in multiple studies, such as the fact that safety attitude has been proven to significantly and positively predict safety behavior, while psychological capital and safety awareness indirectly promote the implementation of safety behavior by enhancing individuals’ emphasis on safety behavior [37,38]. Therefore, safety psychology is not an accessory to cognition but a crucial intermediary link determining whether cognition can be implemented. Based on the above literature analysis, we propose the following hypothesis:

H4: 

Safety psychology plays a mediating role between the safety cognition and safety behavior of construction workers.

Work pressure is a psychological and physiological state of tension caused by unfavorable conditions in the work environment, including heavy workloads, tight time constraints, shift work arrangements, high temperatures, and adverse environments. Existing research has shown that there is a significant negative relationship between work pressure and safety behavior among construction workers [39]. Work pressure not only directly leads to an increase in unsafe behaviors but may also weaken the level of safety behavior by affecting the psychological state of workers [40]. For instance, high work pressure may result in the depletion of psychological resources, the accumulation of negative emotions, and a decline in risk perception ability, which further prompts workers to take quick and simple safety measures or ignore safety regulations [41]. In driving behavior research, it is pointed out that drivers’ limited rational cognition and emotional states significantly influence their behavioral decisions, further supporting the mediating role of cognitive factors in behavior generation [42]. Additionally, according to self-determination theory, positive psychological states such as work engagement can enhance an individual’s cognitive flexibility and behavioral adaptability, thereby promoting the implementation of safety behaviors [43]. These studies collectively suggest that construction workers under high work pressure are more likely to exhibit a decline in safety behavior if their safety psychological resources are insufficient, while enhancing safety psychological resources may buffer the negative impact of stress on safety behavior. Based on the above literature analysis, we propose the following hypothesis:

H5: 

Safety psychology plays a mediating role between work pressure and safety behavior of construction workers.

2.6. Moderating Effect

In the research on the safety behavior of construction workers, safety motivation reflects the tendency of workers to take safety actions due to their recognition of the importance of safety activities or their perception of safety as part of their personal values [44]. Safety psychology, on the other hand, is a comprehensive psychological state formed by individuals in specific work contexts based on their cognition of the organizational safety atmosphere, management support, and their own roles, involving attitudes, beliefs, and emotions towards safety [27,28].

Autonomous motivation is characterized by workers’ genuine willingness to implement safety behaviors from the heart. When construction workers perceive that they can make autonomous decisions regarding safety matters, believe they have the ability to complete safety tasks, and feel supported by colleagues and supervisors, their safety behavior motivation tends to be more internalized. This internal motivation prompts workers to form more persistent and stable safety practices. This internalization process helps cultivate a safety culture, making compliance with safety measures a shared value among the workforce and further reducing the incidence of work-related injuries.

According to SDT, the regulatory types of controlled motivation can be classified into introjected regulation and external regulation [27,45], which refer to the behavioral tendencies of individuals driven by external pressure or rewards and punishments. It significantly affects an individual’s performance in a safe environment. Previous studies have shown that external safety motivation has a positive impact on safety behavior [46]. Kvorning et al. pointed out that external forces play a crucial role in compelling individuals to perform specific behaviors [47]. For instance, if employees mainly comply with safety regulations to avoid penalties or to meet regulatory requirements, their safety behavior often shows mechanical, passive adaptation and low persistence.

Although from a long-term internalization perspective, controlled motivation is less ideal than autonomous motivation, effective external control remains a necessary component of a sustainable safety management system. These control measures provide a basic level of protection and compliance, which is crucial for maintaining a safe working environment and preventing immediate injuries, and also meet the basic requirements of decent work. Based on the above literature analysis, we propose the following hypotheses:

H6: 

Autonomous motivation plays a moderating role between the safety psychology and safety behavior of construction workers.

H7: 

Controlled motivation plays a moderating role between the safety psychology and safety behavior of construction workers.

Social identity is defined as an individual’s awareness or knowledge of belonging to a certain group, as well as the emotional and value significance of this group membership to them. In the context of construction workers’ safety participation, project identity, as a specific form of social identity, is introduced into the research model to explain the workers’ extra-role behaviors in safety management. For instance, workers who have a strong sense of identity with the project they are involved in are more likely to actively participate in safety activities, help colleagues complete safety tasks, and propose safety improvement suggestions [27].

However, the effect of cognition on behavior is not simple and direct, and psychological mediating mechanisms are particularly crucial in this process. Social identity theory indicates that an individual’s sense of belonging to a group can significantly influence the way their cognition is transformed into behavior [48]. Specifically, when workers have a strong sense of identity with the project or team, they are more inclined to internalize the group’s safety norms and have a stronger motivation to convert safety cognition into actual behavior [49,50]. Additionally, research has found that role identity, as a specific form of social identity, can moderate the impact of different types of motivation on safety behavior [44]. For example, in a high-identity state, workers are more likely to sustain safety behaviors driven by autonomous motivation, while in a low-identity state, even if controlled motivation is still at play, the persistence and quality of their behavior will decline. Based on the above literature analysis, we propose the following hypothesis.

H8: 

Social identity moderates the relationship between safety cognition and safety behavior of construction workers.

Figure 1 presents the conceptual framework of this study. The model elucidates the relationships among variables: safety cognition, safety psychology and job stress are hypothesized to directly influence safety behavior (H1–H3), while safety psychology mediates their indirect effects on safety behavior (H4, H5). The model further reveals that social identity moderates the relationship between safety cognition and safety behavior (H8), whereas autonomous and control motivation moderate the relationship between safety psychology and safety behavior (H6, H7).

3. Method

3.1. Sample and Procedures

This study selected multiple key construction sites under large state-owned enterprises such as China Railway and China Construction as sampling objects, adopting a stratified quota sampling design based on industry types and work experience, consistent with the specifications of the “China Construction Industry Labor Report”. Participants were selected from the on-site list through simple random sampling in each stratum, and the representativeness of the samples was verified by comparing with data from the National Bureau of Statistics. The study covered various trades, including concrete workers, steelworkers, and ordinary construction workers, distributing a total of 469 questionnaires, with 376 returned and 334 valid responses obtained, resulting in a questionnaire return rate of 80.17% and an effective sample return rate of 88.83%. Among these, 316 were filled out by grassroots workers and 18 by management personnel. The questionnaire data were statistically analyzed using SPSS 27 and AMOS 28.

As shown in

Table 1. Socio-demographic details of questionnaire respondents.

Items	Details	Frequency	Percent
Gender	Male	290	86.8%
	Female	44	13.2%
Age	18–25	82	24.6%
	26–33	65	19.5%
	34–44	52	15.6%
	45–55	62	18.6%
	≥56	73	21.9%
Education	Primary school and below	78	23.4%
	Junior middle school	87	26.0%
	Senior middle school	89	26.6%
	College, bachelor’s degree and above	80	24.0%
Number of households	1	2	0.6%
	2	11	3.3%
	3	66	19.8%
	4	110	32.9%
	5	76	22.8%
	≥6	69	20.7%
Work experience	≤5 years	55	16.5%
	5–10 years	73	21.9%
	11–15 years	54	16.2%
	16–20 years	88	26.3%
	≥20 years	64	19.2%
Accident experience	0	28	8.4%
	1–3 times	105	31.4%
	4–6 times	98	29.3%
	7–12 times	73	21.9%
	≥13 times	30	9.0%

, males account for 86.8% of the sample, which closely aligns with the gender distribution characteristics in China’s construction industry. Notably, 76% of practitioners hold only high school diplomas or lower educational qualifications, while 61.7% possess over 11 years of work experience. This indicates that although this group has substantial experience, their overall educational level remains relatively low. These demographic features are typical in the field, highlighting the importance of tailoring safety interventions for this specific group—ensuring both accessibility and effectiveness of these measures.

3.2. Measures

3.2.1. Multi-Dimensional Scale Design

To comprehensively measure the safety behavior mechanism of construction workers, this study drew on the classic scales from SDT, including the Intrinsic Motivation Inventory (IMI), and made deletions, modifications, and contextual adjustments to some items based on the specific circumstances of the research subjects to better align with the actual lives and research needs of construction workers. In addition to the mature Western scales, it also integrated the safety motivation scale developed by Chinese scholars to enhance local applicability. Building on this foundation, the study further expanded its measurement dimensions to include core variables such as safety cognition, safety psychology, work pressure, and social identity. To ensure clarity and operational feasibility of key constructs, the research provided explicit definitions and measurement methods: Safety cognition refers to workers’ ability to perceive, understand, and make decisions regarding on-site hazards. This is measured through items like “I know how to complete tasks in a safe manner”. Safety psychology is the comprehensive psychological state of an individual in a specific working environment. It is measured by items, such as “no matter how serious the accident or hidden danger is, I will deal with it in time”. The 5-point Likert scale was used for measurement. Work pressure is defined as subjective perception of stress caused by high-intensity work conditions, assessed through items such as “I look forward to the holiday, but I feel depressed when I think about working on Monday”). Social identity measures workers’ sense of belonging and value identification with their project team, including items such as “the success of this project is my success”. All core indicators were measured using a standardized five-point Likert scale, with scores ranging from 1 (strongly disagree) to 5 (strongly agree). This quantitative assessment system provides a clear basis for the objective analysis of participants’ psychological states.

3.2.2. Safety Motivation

Since the previous safety motivation scales did not reflect their qualitative differences, the study compiled its own safety motivation scale with reference to the universal motivation scale compiled by Koestner et al. [51]. The scale includes 4 dimensions: identified regulation, intrinsic regulation, external regulation and introjected regulation; each dimension has 2 items, and the total table has 8 items. This granular assessment of motivation types is crucial for understanding pathways to foster more sustainable, self-determined safety behaviors.

3.2.3. Safe Behavior

Employees compiled by Vinodkumar et al. [52] were employed for security task behavior and security citizen behavior, respectively. Measuring these safety outcomes provides the empirical basis for evaluating interventions aimed at enhancing sustainable safety performance in construction.

3.3. Data Analysis Procedure

To ensure methodological rigor, the data analysis follows a multi-stage process. Meanwhile, to clearly demonstrate the research methodology of this paper, a research methodology flowchart was created, as shown in Figure 2 below:

4. Analyses and Results

4.1. Common Method Bias Test

The study employed Harman’s single-factor test to assess common method bias, conducting an exploratory factor analysis with all variable items. The results indicated that the first factor (unrotated) accounted for 30.32% of the total variance, below the 50% threshold [53], suggesting no significant common method bias in the data.

4.2. Reliability Analysis

In this study, the primary factors were assessed using standardized scales, making the evaluation of data quality essential for ensuring meaningful analysis. As an initial analytical step, Cronbach’s alpha was utilized to evaluate the internal consistency across all measurement dimensions. The reliability coefficients derived from this analysis ranged between 0 and 1, with scores approaching 1 denoting enhanced internal consistency within the assessed constructs. A reliability coefficient below 0.6 is typically deemed unreliable, prompting a need to redesign the questionnaire or to recollect and reanalyze the data. Coefficients from 0.6 to 0.7 are considered acceptable, those from 0.7 to 0.8 are deemed moderately reliable, values between 0.8 and 0.9 are classified as highly reliable, and those from 0.9 to 1 are considered to be extremely reliable.

As shown in

Table 2. Reliability analysis.

Variable	Number	Cronbach’s Alpha	Overall Cronbach’s Alpha
SC	3	0.778	0.843
SP	3	0.778
WP	3	0.794
SI	3	0.786
AM	3	0.770
CM	3	0.788
SB	4	0.788

, the reliability coefficients of the overall scales and each secondary dimension in this analysis are all within the range of 0.7 to 1, indicating that the scale used in this study has good internal consistency and reliability.

4.3. Validity Analysis

To evaluate the convergent validity (AVE) and composite reliability (CR) of each dimension, the standardized factor loadings for each measurement item within their respective dimensions are first calculated using the established Confirmatory Factor Analysis (CFA) model. These calculations provide the basis for determining the AVE and CR values for each dimension by applying the respective formulas for AVE and CR. In accordance with accepted standards, an AVE value of at least 0.5 and a CR value of at least 0.7 are required to confirm adequate levels of convergent validity and composite reliability.

The results, as presented in

Table 3. Testing Convergent Validity and Composite Reliability.

Path Relationships			Estimate	AVE	CR
SC3	←	SC	0.68	0.609	0.822
SC2	←	SC	0.81
SC1	←	SC	0.74
SP3	←	SP	0.8	0.602	0.819
SP2	←	SP	0.78
SP1	←	SP	0.67
WP3	←	WP	0.66	0.616	0.828
WP2	←	WP	0.76
WP1	←	WP	0.81
SB4	←	SB	0.79	0.609	0.862
SB3	←	SB	073
SB2	←	SB	0.73
SB1	←	SB	0.78

, show that the AVE values for all dimensions of the health maintenance scale are greater than 0.5, while the CR values exceed the threshold of 0.7. These findings provide strong evidence that each dimension possesses good convergent validity and composite reliability, meeting the required standards.

To evaluate the psychometric validity of the measurement scale, an exploratory factor analysis (EFA) was performed within the SPSS 27 statistical software environment. The principal component analysis (PCA) technique was selected as the extraction method to reduce dimensionality, followed by the application of a varimax orthogonal rotation. Factors were retained based on eigenvalues greater than 1, and factor loadings were organized in descending order. The results of this validity evaluation through EFA are presented in

Table 4. KMO and Bartlett’s Test.

Kaiser–Meyer–Olkin Measure of Sampling Adequacy		0.846
Bartlett’s test of sphericity	Approx. chi-square	2266.023
	Df	153
	Sig.	0.000

. The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy exceeds the commonly accepted threshold of 0.8, indicating a high level of suitability for factor analysis.

As presented in Table 4, the KMO measure of sampling adequacy was 0.846, exceeding the recommended threshold of 0.6, and Bartlett’s test of sphericity was significant (χ² = 2266.023, p < 0.001). These results confirm that the data are suitable for factor analysis.

4.4. Descriptive Statistics and Normality Tests

Table 5. Descriptive statistics and normality tests.

Construct	Item	Mean	SD	Skewness	Kurtosis	Overall Mean	Overall Standard Deviation
SC	SC1	3.63	0.903	−0.445	0.206	3.4970	0.90646
	SC2	3.57	0.830	−0.208	−0.179
	SC3	3.28	0.891	−0.267	−0.130
SP	SP1	3.47	0.932	−0.134	−0.390	3.4172	0.91038
	SP2	3.44	0.887	−0.042	−0.392
	SP3	3.34	0.909	−0.052	−0.272
WP	WP1	3.76	0.963	−0.580	0.116	3.5499	0.99147
	WP2	3.40	0.933	−0.217	−0.128
	WP3	3.49	1.042	−0.329	−0.493
SI	SI1	3.58	0.982	−0.217	−0.482	3.7026	0.93194
	SI2	3.60	0.923	−0.264	−0.346
	SI3	3.92	0.850	−0.405	−0.344
AM	AM1	3.41	1.020	−0.199	−0.538	3.3503	0.98340
	AM2	3.11	1.009	0.147	−0.560
	AM3	3.53	0.868	−0.130	−0.130
CM	CM1	3.50	0.961	−0.192	−0.426	3.5099	0.97234
	CM2	3.48	0.961	−0.104	−0.524
	CM3	3.55	0.997	−0.279	−0480
SB	SB1	3.28	0.925	−0.207	−0.223	3.4326	0.93293
	SB2	3.49	0.916	−0.215	−0.304
	SB3	3.26	0.923	−0.096	−0.547
	SB4	3.71	0.898	−0.353	−0.244

presents descriptive statistical analysis and normality testing of the factors in use. Initially, the descriptive statistics revealed that the mean scores for each variable ranged from 3 to 4. Considering the scale’s scoring method ranges from 1 to 5 with positive scoring, these results indicate that the study population has at least an average level of awareness and behavior concerning safety. Further, the normality of the measurement items was tested using skewness and kurtosis indices. According to statistical standards, data is considered to approximate a normal distribution if the absolute values of skewness do not exceed 3 and those of kurtosis do not exceed 8.

Table 5 shows that all item means are above the midpoint of 3, indicating generally positive responses. The skewness and kurtosis values for all items are within the acceptable range, suggesting that the data approximate a normal distribution.

4.5. Correlation Analysis

To investigate potential associations among the theoretical variables, Pearson’s correlation coefficient analysis was performed as a primary statistical method in this study, as shown in

Table 6. Pearson correlation analysis results among dimensions.

Dimension	SC	SP	WP	SI	AM	CM	SB
SC	1
SP	0.320 **	1
WP	0.375 **	0.378 **	1
SI	0.306 **	0.336 **	0.327 **	1
AM	0.284 **	0.337 **	0.278 **	0.325 **	1
CM	0.321 **	0.356 **	0.242 **	0.281 **	0.534 **	1
SB	0.331 **	0.385 **	0.349 **	0.388 **	0.326 **	0.271 **	1

Note: ** Correlation is significant at the 0.01 level.

. This approach facilitates the quantification of linear relationships between paired variables, adhering to parametric assumptions of normality and scale continuity inherent to the dataset.

The correlation matrix in Table 6 indicates that all variables are significantly and positively correlated with each other at the 0.01 level. Notably, SB has significant positive correlations with all predictor variables, providing preliminary support for our hypothesized relationships.

4.6. Structural Model Analysis and Hypothesis Tests

SEM analysis based on 334 valid questionnaires showed that the sample size met the minimum standard requirements, and all model goodness-of-fit indices fully complied with the criteria. The model fit indices presented in

Table 7. Model fit test.

	CMIN/DF	RMSEA	IFI	NFI	CFI
Recommended Standards	<3	<0.08	>0.9	>0.9	>0.9
Actual Measurement Results	1.596	0.042	0.961	0.952	0.961

provide strong evidence for the adequacy of the CFA model used in this study. The CMIN/DF (chi-square to degrees of freedom ratio) was calculated at 1.596, which falls comfortably within the generally accepted range of 1 to 3, suggesting a reasonable fit between the hypothesized model and the observed data. The RMSEA (Root Mean Square Error of Approximation) is 0.042, which is well below the maximum acceptable threshold of 0.08, further indicating a good fit. Additionally, the Incremental Fit Index (IFI), Normed Fit Index (NFI), and Comparative Fit Index (CFI) all exceed the critical value of 0.9, which is commonly recognized as the benchmark for an excellent model fit. Each of these indices demonstrates that the proposed model closely aligns with the observed data, satisfying both absolute and incremental fit criteria.

Based on the good fit indices mentioned above, the path analysis results of the structural equation model are presented in Figure 3, which displays all the standardized path coefficients and their significance levels.

The results of the hypothesis testing, including the path coefficients and their corresponding p-values, are summarized in

Table 8. Results of hypothesis testing.

Hypothesis Path	S.E.	C.R.	p-Values
SC → SP	0.087	2.566	0.01
WP → SP	0.085	4.621	<0.001
SP → SB	0.073	4.149	<0.001
SC → SB	0.082	2.543	0.011
WP → SB	0.082	2.864	0.004

, and the results of the moderation effects are summarized in

Table 9. Results of the moderation effect analysis.

	Model 1	Model 2	Model 3 (SI)	Model 3 (AM)	Model 3 (CM)
Constant	3.433 *** (87.946)	3.433 ***	3.396 *** (89.544)	3.405 *** (87.917)	3.399 *** (85.671)
SC	0.336 *** (6.394)	0.238 *** (4.541)	0.246 *** (4.776)	0.277 *** (5.253)	0.289 *** (5.312)
SP	0.379 *** (7.596)	0.306 *** (5.910)
SI		0.313 *** (6.122)	0.327 *** (6.470)
AM		0.206 *** (4.208)		0.219 *** (4.492)
CM					0.160 ** (3.212)
SC × SI			0.212 *** (3.482)
SP × AM				0.133 * (2.415)
SP × CM					0.133 * (2.415)
Sample size	334	334	334	334	334

* p < 0.05, ** p < 0.01, *** p < 0.001, inside the parentheses is the value of t.

. According to the data in Table 8 and Table 9, all p-values are below 0.05, indicating that each hypothesis has been verified at the significance level of 0.05.

This study employed the maximum likelihood estimation method for path analysis. The model test results are detailed in Table 9. The analysis indicates that safety cognition and work pressure have significant direct impacts on safety behavior, with path coefficients reaching statistical significance. More importantly, safety psychology plays a significant mediating role between safety cognition, work pressure, and safety behavior, suggesting that the internal psychological state of workers is a key mechanism for transmitting the influence of cognition and pressure. Additionally, social identity significantly moderates the effect of safety cognition on safety behavior, while autonomous and controlled motivation have moderating effects on the transformation path of safety psychology. These findings not only validate the core path of “cognition–psychology–behavior”, but also reveal the complex mechanism of the formation of construction workers’ safety behavior from multiple levels.

Based on the above analysis, PROCESS 4.1 was further employed to test the moderating effects of social identity on the relationship between safety cognition and safety behavior, as well as the moderating effects of autonomous and controlled safety motivation on the relationship between safety psychology and safety behavior. The study comprehensively examined the interactive effects of SC, SP, SI, AM, and CM on SB. In Model 1, SC and SP had significant positive effects on SB, with path coefficients of 0.336 and 0.379 (p < 0.001), respectively. After introducing SI, AM, and CM as moderating variables, the results of Model 2 and Model 3 showed that SI (path coefficient 0.313, p < 0.001) and AM and CM all exhibited significant moderating effects. Specifically, SI significantly enhanced the promoting effect of SC on safety behavior (Int_1 = 0.212, p < 0.001); AM also played a positive moderating role between SP and SB (Int_1 = 0.133, p = 0.05); and the moderating effect of CM on the relationship between SP and SB was also significant (Int_1 = 0.153, p = 0.01), indicating that external incentives or supervisory factors may further regulate the intensity of the impact of SP on SB. To present the above moderating effects more intuitively, we further drew simple slope graphs for analysis. Figure 4, Figure 5 and Figure 6 show that under high SI conditions, the predictive slope of SC on SB is steeper; similarly, high AM also enhances the positive impact of SP on SB; the moderating effect of CM is also reflected in the figures, demonstrating the significant differentiation of the relationship between SP and SB under different motivation levels. The R² values of all models were significantly improved, indicating that the introduction of the above moderating variables effectively enhanced the model’s explanatory power for safety behavior, reflecting the complex mechanism by which internal cognitive and psychological mechanisms and external social and motivational factors jointly influence the formation of safety behavior.

In conclusion, all hypotheses mentioned above have been verified. These empirical results provide robust evidence for the proposed SDT-based model, offering valuable insights for developing psychologically informed strategies to enhance construction safety, a critical component of social sustainability and decent work.

4.7. Mediation Analysis

To explore the internal mechanism by which safety psychology has a significant positive impact on safety behavior, this study introduced safety psychology as a mediating variable in the structural equation model. Through the bias-corrected Bootstrap method, the mediating effect was tested, verifying the mediating role of safety psychology between safety cognition, work pressure, and safety behavior to form a chain mediating model. The results of the mediation effect are shown in

Table 10. Test of Mediating Effect.

Intermediary Path	Estimate	Lower	Upper	p
SC-SP-SB	0.067	0.014	0.146	0.014
WP-SP-SB	0.118	0.051	0.207	0.001

.

The results in Table 10 indicate that the mediating role of SP between SC and SB, as well as between WP and SB, has reached a statistically significant level. Specifically, the mediating effect value of the path “safety cognition–safety psychology–safety behavior” is 0.067, with a 95% Bootstrap confidence interval of [0.014, 0.146], which does not include 0 (p = 0.014); the mediating effect value of the path “work pressure–safety psychology–safety behavior” is 0.118, with a 95% confidence interval of [0.051, 0.207], also not including 0 (p = 0.001). The results confirm that safety psychology is a key mediating variable connecting safety cognition, work pressure, and safety behavior, indicating that the internal psychological state of workers plays a crucial role in the transmission process of external cognitive factors and pressure factors influencing safety behavior. This path is crucial for designing sustainable intervention measures—by fostering an internal commitment to safety behavior, it not only supports the long-term improvement of workers’ health and safety levels but also helps ensure the sustainability of construction operations.

The model confirms that by enhancing workers’ safety cognition, optimizing work pressure management, and strengthening their safety psychological state, it can effectively promote the active internalization and long-term maintenance of safety behaviors. This internal behavioral transformation mechanism reduces the reliance on single external supervision, lowers management costs and accident risks, and directly contributes to the protection of human resources and the improvement of operational reliability during the construction process. From the perspective of sustainable development, ensuring the physical and mental health of workers and reducing occupational injuries are important foundations for achieving inclusiveness and fair employment in the social dimension; at the same time, the continuous implementation of safety behaviors helps to reduce project delays and resource waste caused by accidents and enhances project economic benefits and environmental compatibility. Therefore, incorporating psychologically oriented safety management strategies into the framework of sustainable development not only strengthens corporate social responsibility and worker well-being but also provides a theoretical basis and practical paths for building a more resilient and human-oriented construction production system.

5. Discussion

5.1. Theoretical Implications

This study integrates the perspectives of cognitive psychology, motivation theory, and social identity theory to construct a theoretical model that explains the formation mechanism of safety behaviors among construction workers. The empirical test conducted in the local context of China has yielded numerous theoretical achievements.

This study breaks through the traditional research paradigm in safety management. The findings demonstrate that external cognitive stressors can be transformed into behavioral manifestations through changes in internal psychological states, providing a more nuanced theoretical explanation for the formation of safety behaviors. For instance, at construction sites, when workers encounter safety knowledge, their psychological states, such as safety attitudes or emotional responses, directly influence behavioral decisions, thereby transcending the limitations of simplistic cause-and-effect relationships.

Secondly, research indicates that intrinsic motivation originates from personal values and interests, emphasizing the reinforcement of safety habits through self-motivation and long-term commitment, which exerts integrated and sustained influence on safety behaviors. Conversely, extrinsic motivation primarily relies on immediate constraints such as external rules and reward–punishment mechanisms, for example, rapidly adjusting behaviors through supervision and institutional pressure. This study innovatively employs intrinsic and extrinsic motivation as moderating variables to indirectly investigate their impact on construction workers’ safety performance. The significant moderating effects of autonomous and controlled motivation not only validate the practical application value of workplace safety analysis in high-risk construction environments but also highlight the necessity of dual-motivation strategies in safety management. These findings hold important theoretical significance, demonstrating that motivation types are not merely background factors but dynamic variables that actively regulate the psychological–behavioral relationship.

This study integrates social identity theory into a comprehensive model based on SDT, revealing that strong social identity significantly enhances the conversion efficiency from safety cognition to safety behavior. These findings provide empirical support for applying social identity theory in safety management, demonstrating that group belongingness serves as a crucial psychosocial factor in promoting the internalization of safety norms. The significant moderating effect of social identity highlights its pivotal role in bridging the gap between safety cognition and actual behavior. This discovery holds particular significance as it uncovers a long-overlooked leverage point for safety interventions within traditional safety management frameworks.

This study incorporates cognitive, pressure, psychological, motivational, and social factors into the same analytical framework, providing a multi-level and multi-path theoretical model that can more comprehensively explain the formation mechanism of safety behavior among Chinese construction workers. This model not only has strong explanatory power but also provides a foundation for subsequent cross-cultural and cross-industry comparative studies.

5.2. Practical Implications

The results of this study provide systematic and operational practical implications for construction enterprises, safety managers, and policymakers, which can help improve the safety level of construction sites from multiple dimensions.

Enterprises should fundamentally strengthen safety psychological construction and incorporate psychological empowerment content into regular safety training, such as through scenario simulation and case review, to enhance workers’ sensitivity to risks and confidence in their ability to respond. Meanwhile, a “dual-track drive” incentive mechanism should be implemented. This approach not only enhances immediate safety performance but also cultivates a sustainable workforce with high morale and low turnover rates. Additionally, companies should focus on building project culture and strengthening team cohesion to enhance employees’ sense of social belonging and organizational identification. This fosters voluntary compliance with safety protocols. Consequently, reduced accident rates will lower waste disposal costs and resource consumption, effectively improving both the economic and environmental benefits of the project.

For safety managers, they should adopt more refined incentive strategies, identify the motivation types of different employees, and implement differentiated guidance. For instance, for workers with strong autonomous motivation, more on-site decision-making power can be appropriately granted, while for those who rely more on external constraints, clear reward and punishment terms should be maintained to strengthen behavioral orientation. Managers should also promote the establishment of a regular psychological support mechanism, through regular assessment and timely intervention, to alleviate the consumption of cognitive resources caused by high pressure on workers and prevent safety negligence due to psychological fatigue.

At both industry and policy levels, we recommend implementing an innovative safety training model that integrates intelligent technologies with psychological insights. By combining immersive technologies like VR and AR with psychological assessment modules, this approach can significantly enhance training effectiveness and employee engagement. Such an integrated system would systematically guide the industry to move beyond a narrow focus on lagging indicators, shifting toward a sustainability-oriented, holistic safety paradigm that positions human capital as the core asset for long-term survival.

This study establishes a micro-level foundation for sustainable development in the construction industry by leveraging the internalization of safety behaviors through SDT-based interventions. The reduction in accidents directly curbs material and energy waste caused by emergency response and rework, thereby advancing environmental sustainability. Simultaneously, decreased workplace injuries and fatalities not only preserve human capital and alleviate public health system burdens but also prevent families from falling into poverty due to income loss—key elements of social sustainability. From an economic perspective, safer working environments enhance productivity and reduce costs, driving economic sustainability in project outcomes and business operations. Therefore, investing in psychological mechanisms that support safety behaviors is not merely a compliance issue but a strategic imperative for building a safer, more efficient, and truly sustainable construction industry.

5.3. Limitations and Future Research

5.3.1. Limitations

Although this study has accumulated valuable experience, several limitations need to be particularly noted. First, the research sample was only from construction workers in a single country, which may limit the generalizability of the research conclusions to other regions or industries. The use of a Chinese sample in our study restricted the universal applicability of the research results. Second, although this study focused on the impact of cognition, psychology, and stress on safety behavior, it must be acknowledged that other factors may also affect the safety behavior of construction workers, such as organizational culture, leadership style, or task requirements. Future research can further examine these variables to gain a more comprehensive understanding of safety behavior in the construction industry. Finally, the cross-sectional design adopted in this study limited the establishment of causal relationships between variables. It is recommended that future studies use longitudinal or experimental designs to explore the causal relationships between cognition, psychology, and safety behavior. Longitudinal studies are crucial for understanding the long-term dynamic changes and sustainability of safety improvement measures driven by incentives.

5.3.2. Future Research

Based on these limitations, future research can proceed in multiple directions: First, it can explore the impact of individual differences among construction workers (such as age, education level, work experience, personality traits, etc.) on the relationship between cognition, psychology, and safety behavior, thereby providing important clues for predicting individual differences in safety behavior. Second, given the dynamic nature of safety behavior in the workplace, it is necessary to conduct longitudinal studies tracking changes in safety behavior and motivation to explore the patterns of these behaviors over time. In addition, research on how to utilize digital technology to meet psychological needs and cultivate autonomous safety motivation will open up a promising new path for enhancing sustainable safety management.

6. Conclusions

This study, grounded in Self-Determination Theory, provides a clear quantitative framework for construction safety management through empirical analysis. The findings reveal that stress reduction emerges as the most effective intervention —its mediating effect on safety behavior through psychological factors (β = 0.118) demonstrates approximately 76% greater efficacy than the direct pathway through safety cognition (β = 0.067), indicating that stress management yields greater safety outcomes than knowledge training alone. Social identity emerges as the optimal “catalyst” for behavioral transformation, with its moderating effect (β = 0.212) significantly surpassing motivational factors, suggesting that team cohesion proves more effective than incentive mechanisms in translating safety knowledge into action. Notably, the moderating effects of autonomy and control motivation are identical (β = 0.133), confirming the critical importance of a dual-track” strategy combining internal drive and external constraints.

The accurate analysis of psychological mechanisms underlying construction workers’ safety behaviors provides both theoretical and empirical support for addressing the persistent challenge of high accident rates in the construction industry. The study deepens theoretical understanding of how three psychological needs—autonomy, competence, and belonging—shape safety behaviors. Practical recommendations include three strategies: establishing worker safety committees, implementing VR/AR training programs, and launching safety partnership initiatives. By addressing workers’ psychological needs, these measures effectively stimulate autonomous safety behaviors. Policy recommendations suggest incorporating psychological indicators into industry standards and establishing dedicated support funds to continuously optimize safety management models. Implementation of this mechanism will significantly reduce workplace accident rates and occupational disease incidence, alleviate public healthcare burdens, and fundamentally eliminate poverty risks caused by work-related injuries. This provides scientific evidence for transforming the construction industry into a safer, more humane, and more sustainable sector.