Research on the Influence of Anxiety Psychology on Unsafe Behavior Among Construction Workers

Abstract

Unsafe behaviors among construction workers constitute a primary cause of safety accidents, with unsafe psychological states often triggering such behaviors. Focusing on collective anxiety issues, this study introduces the standardized anxiety scale (SAS) into the construction domain to investigate the influence mechanism of anxiety psychology on construction workers’ unsafe behaviors. A hypothesized model was established, incorporating organizational climate, safety competency, and job satisfaction as mediating variables, and demographic characteristics (gender, age, educational background, work experience, anxiety level) as moderating variables. Data collected from field surveys were analyzed using SPSS 22.0 and AMOS 24.0 for correlation analysis, variance analysis, and regression analysis, while structural equation modeling (SEM) was employed to validate the theoretical model and hypotheses. The results indicate that (1) anxiety psychology exhibits a significant positive correlation with unsafe behaviors; (2) organizational climate, safety competency, and job satisfaction mediate the transmission pathways between anxiety psychology and unsafe behaviors; (3) enhancing organizational climate, improving safety competency, and ensuring job satisfaction effectively reduce safety incident rates. Additionally, this study proposes preventive measures targeting anxiety psychology across three dimensions: external environmental controls, internal self-regulation, and direct anxiety intervention. These measures provide novel perspectives for effectively reducing the occurrence of unsafe behaviors among construction workers and advancing safety governance frameworks in the construction industry.

1. Introduction

As a pivotal pillar of China’s national economy, the construction industry has consistently contributed over 6.5% to GDP growth in the past decade, demonstrating significant socioeconomic value. However, the labor-intensive nature, complex working environments, and technological diversity of this sector have led to persistent safety challenges. From 2018 to 2022, China’s construction sector recorded 4198 fatal accidents, with an average daily death toll of 2.3 and direct economic losses exceeding CNY 10 billion, starkly exposing systemic safety deficiencies. Studies indicate that approximately 80% of safety incidents originate from workers’ unsafe behaviors [1], which are closely tied to psychological states—particularly anxiety psychology [2,3]. Recent years have witnessed escalating societal anxiety (including employment anxiety [4], health anxiety [5], etc.), with construction workers being disproportionately vulnerable due to high-intensity labor, hazardous working conditions, and social isolation [6]. These stressors exacerbate anxiety, triggering unsafe acts and perpetuating a vicious cycle of risk amplification. Therefore, elucidating the mechanisms through which anxiety psychology influences unsafe behaviors holds urgent practical significance for enhancing safety management practices in the construction industry.

In construction safety research, accident causation is generally attributed to the interplay of four factors—human, material, environmental, and managerial elements [7], with human factors being particularly emphasized. Scholars have extensively explored unsafe behaviors from a human-centric perspective, demonstrating that psychological processes critically determine their occurrence [8]. Chen, W et al. [9] investigated unsafe psychology–behavior linkages among construction workers, confirming strong correlations between unsafe acts and psychological states. Mei, C et al. [10] revealed that subjective social support (SSS) enhances workers’ subjective well-being, mitigates paranoid tendencies, and improves safety awareness and emergency response capabilities, thereby reducing unsafe behaviors and injury risks. Ma, H et al. [11], applying the Theory of Planned Behavior (TPB), utilized structural equation modeling (SEM) to analyze how optimistic bias moderates ambivalence in deliberate unsafe behavior decision-making, identifying complacency and paralytic negligence as key drivers of safety incidents. Saleem et al. [12] investigated the impact of psychological capital (PsyCap) on construction workers’ safety behaviors. Their research demonstrates that PsyCap enhances work engagement and promotes safety behaviors by cultivating positive psychological resources such as self-efficacy, hope, optimism, and resilience, thereby effectively reducing the likelihood of unsafe behaviors. Collectively, existing studies focus on either individual characteristics or macro-level psychological effects, leaving the specific mechanisms linking anxiety psychology to unsafe behaviors underexplored.

The novelty of this research resides in the development of a specialized measurement scale to assess the interactive dynamics between construction workers’ anxiety psychology and their manifested unsafe behaviors, along with the establishment of a theoretical framework delineating the anxiety psychology–unsafe behavior relationship. Structural equation modeling (SEM) was employed to validate and explicate the mechanisms through which anxiety psychology, organizational climate, safety competency, job satisfaction, and individual characteristics collectively influence unsafe behaviors. Furthermore, grounded in a sociocognitive perspective, this research proposes targeted intervention measures addressing anxiety-related unsafe behaviors, offering innovative directions and actionable insights for enhancing project safety performance through integrated psychological and organizational strategies.

2. Literature Review and Theoretical Model

2.1. Unsafe Behaviors and Unsafe Psychology

The Theory of Planned Behavior (TPB), one of the most prominent frameworks in behavioral psychology, investigates the latent mechanisms underlying individual behaviors. According to TPB, unsafe behavior intention refers to an individual’s decision-making state when evaluating whether to engage in unsafe acts in real-world contexts, constituting the cognitive foundation for such behaviors [13]. Unsafe behaviors remain a primary cause of construction accidents, with their influencing factors and prevention mechanisms being persistent research foci. Scholars have extensively explored determinants across individual, organizational, and environmental dimensions, identifying critical factors such as safety awareness, on-site management, and external conditions [14,15]. At the individual level, analyses emphasize workers’ physiological and psychological states [16]. Psychological factors significantly drive behaviors, with research indicating that unsafe psychology—manifested as complacency [17], paralysis [16], and conformity [18]—acts as an intrinsic motivator for unsafe acts. Wang, D et al. [19] demonstrated that psychological adjustment alleviates safety-related stress, enhances worker engagement, and reduces unsafe behaviors. Zhang et al. [20] integrated the Big Five personality model and TPB to examine how psychological variables—particularly risk propensity, personality traits, and behavioral intentions—influence construction workers’ unsafe acts. Their findings revealed significant associations among personality traits, risk propensity, and unsafe behavior intentions. These studies collectively suggest that psychological resilience directly modulates individuals’ physiological and psychological responses under stress. Excessive stress induces physiological fatigue and emotional instability [21], impairing workers’ psychophysiological functioning and compromising their capacity to maintain safe practices, thereby escalating unsafe behaviors.

2.2. Anxiety Psychology and Unsafe Behaviors

In psychology, anxiety is primarily conceptualized as a psychological state characterized by unpleasant emotions such as tension, unease, and distress in anticipation of potential threats [22]. Early research on anxiety psychology centered on psychoanalytic, cognitive–behavioral, and humanistic schools. For instance, Sullivan [23] posited that anxiety arises from interpersonal relationships, with social interactions inherently triggering mild anxiety in most individuals. Higgins [24] proposed the self-discrepancy theory, emphasizing cognitive conflict as a driver of anxiety. Rollo May posited that human development necessitates anxiety, as it generates psychological tension that activates individuals’ adaptive response mechanisms. This tension mobilizes cognitive resources and latent potential, enabling proactive problem-solving in challenging situations [25].

Contemporary studies predominantly focus on macro-level analyses of societal anxiety. Several scholars have investigated direct triggers of miners’ unsafe behaviors, identifying anxiety, fear, depression, and boredom as key emotional factors influencing unsafe psychology [26,27,28]. Anxiety impacts production safety primarily through cognitive biases (e.g., prioritizing immediate concerns over long-term benefits) and behavioral pattern alterations. Short-term anxiety triggers the secretion of stress hormones (e.g., cortisol, adrenaline), enhancing neuromuscular responsiveness and cognitive acuity to cope with sudden workloads. However, chronic anxiety disrupts hormonal balance, impairing cardiopulmonary and neurological functions, leading to uncontrollable worry (e.g., rapid breathing, heart palpitations) and escalating safety risks [29]. To mitigate anxiety’s adverse effects, cognitive behavioral therapy (CBT) has been validated as an effective intervention for health-related anxiety, alleviating symptoms of depression and anxiety [30]. Additionally, short-term mindfulness meditation training not only reframes negative cognitive schemas about anxiety but also enhances metacognitive control, corrects cognitive distortions, and improves adaptive coping strategies [31].

Amid rapid societal development, the convergence and clash of multicultural dynamics, conflicts between traditional and modern values during social transitions, and escalating life and employment pressures have collectively contributed to widespread anxiety [32]. Construction workers, exposed to high-intensity labor, hazardous environments, and social isolation, are particularly susceptible to anxiety. Although high-pressure work environments and societal anxiety may exacerbate anxiety levels, the interaction mechanisms between these contextual factors and unsafe behaviors remain unclear. Specifically, the precise pathways through which anxiety psychology influences unsafe behaviors in construction workers—the core workforce—are yet to be elucidated. This underscores the need to systematically investigate which factors shape construction workers’ anxiety and how such anxiety translates into concrete unsafe behaviors.

Based on the above analysis, the following hypothesis is proposed:

H1: 

Construction workers’ anxiety psychology is positively correlated with unsafe behaviors.

2.3. Mediating Roles of Organizational Climate, Safety Competency, and Job Satisfaction

The organizational environment constitutes the primary context for construction workers’ long-term engagement, where heightened organizational emphasis on safety—manifested through standardized safety education and robust safety protocols—correlates with increased worker safety awareness and reduced accident incidence [33]. Organizational climate exhibits a direct predictive relationship with safety behaviors, serving as one of the most critical antecedents of safe practices. Hang, Z et al. [34] developed a theoretical framework integrating interpersonal relationship theory and group dynamics, demonstrating that group safety climate significantly enhances workers’ safety behaviors through SEM validation. Cheng, J et al. [35] conceptualized organizational climate, unsafe motivations, safety attitudes, safety competency, and unsafe behaviors as an interconnected system, employing structural equation modeling (SEM) and DEMATEL (Decision Making Trial and Evaluation Laboratory) methods to explore unsafe behavior mechanisms. Their findings delineate the pathways through which organizational climate influences safety behavior norms and propose targeted countermeasures. He, C et al. [36] analyzed data from 119 supervisors and 536 workers across 22 Chinese construction projects using SEM, revealing that safety behaviors partially mediate the impact of safety climate on safety outcomes, while highlighting the necessity of prioritizing supervisors’ mental health under high-pressure conditions. Social exchange theory posits that the organization–employee relationship operates on reciprocal exchanges: organizations provide supportive work environments, and employees reciprocate through productive contributions [37]. Discrepancies between organizational climate and worker expectations, however, may yield detrimental outcomes, undermining both safety performance and psychological well-being.

Based on the above analysis, organizational climate is hypothesized to mediate the relationship between anxiety psychology and unsafe behaviors:

H2: 

Organizational climate is negatively correlated with construction workers’ unsafe behaviors.

H3: 

Organizational climate is negatively correlated with construction workers’ anxiety psychology.

H4: 

Organizational climate mediates the relationship between anxiety psychology and unsafe behaviors.

Safety competency encompasses the knowledge, cognitive abilities, technical skills, and physical fitness that enable workers to mitigate safety risks. The construction industry’s complexity, high hazards, intense workloads, and prolonged operations necessitate strong safety competencies. Chen, N et al. [38] classified safety competency into two dimensions—safety behavioral capacity and safety literacy—based on an analysis of construction workers’ safety performance requirements. Broadly, safety competency encompasses three domains: safety cognition, job skills, and physical fitness. Research further indicates that safety competency—defined as the capacity of organizations or individuals to govern risks through intrinsic attributes such as knowledge, experience, skills, value orientations, and motivational drivers—requires systematic enhancement across three dimensions: safety knowledge acquisition, experiential learning, and skill mastery [39]. Within the construction industry’s context of high-risk operations, complex environments, and intensive labor demands, this tripartite competency development framework is essential to meet workers’ heightened safety requirements when addressing hazardous scenarios. Li, S et al. [40] conceptualize safety competency as a skill set integrating knowledge, expertise, and experience essential for safe task execution, which significantly influences safety behaviors through multifaceted mediating effects across structural and cognitive dimensions. Social cognitive theory underscores human agency, positing that individual traits critically shape behaviors through dynamic interactions between personal perceptions and environmental contexts. This theoretical lens emphasizes that workers’ behavioral responses are adaptive syntheses of internal states and external situational demands.

Based on the above analysis, safety competency is hypothesized as follows:

H5: 

Safety competency is negatively correlated with construction workers’ unsafe behaviors.

H6: 

Safety competency is negatively correlated with construction workers’ anxiety psychology.

H7: 

Safety competency mediates the relationship between anxiety psychology and unsafe behaviors.

Job satisfaction is generally defined as an individual’s psychological state resulting from their productive activities within an organizational context, a construct widely utilized in psychological research. As a key metric for assessing productivity and operational efficiency in the construction industry, construction workers’ job satisfaction profoundly influences organizational behavior optimization [41]. Studies indicate that achieving higher job satisfaction necessitates cultivating work environments characterized by competitive compensation, robust organizational citizenship behaviors, and low turnover rates [42]. Ni, G et al. [43] demonstrated that job satisfaction significantly corrects safety non-compliance and safety non-participation among construction workers, with work engagement fully mediating these relationships. Enhancing job satisfaction thus serves as a behavioral corrective mechanism for unsafe acts. Xie, L et al. [44], employing structural equation modeling (SEM), validated the impact of psychological safety climate (PSC) on workers’ intent to stay and the mediating role of job satisfaction in Chinese contexts, using Guangzhou construction workers as a case study. Herzberg’s Two-Factor Theory categorizes determinants of job satisfaction into hygiene factors (e.g., salary, working conditions) and motivational factors (e.g., recognition, career growth). Hygiene factors are essential for maintaining baseline employee morale; their deficiency triggers work withdrawal behaviors, whereas motivational factors must be strategically leveraged to elevate satisfaction levels.

Accordingly, the hypotheses for job satisfaction are proposed as follows:

H8: 

Job satisfaction is negatively correlated with construction workers’ unsafe behaviors.

H9: 

Job satisfaction is negatively correlated with construction workers’ anxiety psychology.

H10: 

Job satisfaction mediates the relationship between anxiety psychology and unsafe behaviors.

2.4. Moderating Effects of Demographic Characteristics

Demographic characteristics refer to quantifiable attributes of individuals, such as gender, age, and educational background, which empirical studies have shown to significantly influence behavioral patterns and psychological states in occupational settings [45]. Variations in research populations and methodological approaches have led to divergent conclusions. Differences in gender, education, career age, working hours, and anxiety levels may result in heterogeneous psychological responses and behavioral tendencies under identical environmental conditions [45,46].

Based on this theoretical framework, the following hypotheses are proposed:

H11: 

Demographic characteristics significantly moderate the manifestations of anxiety psychology and unsafe behaviors among construction workers.

H1a: 

Gender demonstrates statistically significant differences in the manifestations of anxiety psychology and unsafe behaviors.

H1b: 

Age demonstrates statistically significant differences in the manifestations of anxiety psychology and unsafe behaviors.

H1c: 

Educational background demonstrates statistically significant differences in the manifestations of anxiety psychology and unsafe behaviors.

H1d: 

Work experience demonstrates statistically significant differences in the manifestations of anxiety psychology and unsafe behaviors.

H1e: 

Anxiety levels demonstrate statistically significant differences in the manifestations of anxiety psychology and unsafe behaviors.

As shown in Figure 1, integrating the above analyses, a theoretical model can be constructed to elucidate the mechanism through which anxiety psychology influences unsafe behaviors.

3. Methodology

3.1. Research Framework and Theoretical Foundations

The research methodology of this study is shown in Figure 2. To investigate the relationship between anxiety psychology and unsafe behaviors among construction workers, this study adopted three primary methodologies: (a) literature analysis; (b) questionnaire surveys; (c) statistical analysis. Grounded in theoretical and empirical literature, the research framework involved three sequential phases: first, constructing a hypothesized model of unsafe behavior determinants; second, developing measurement scales for anxiety psychology and unsafe behaviors based on the standardized anxiety scale (SAS); and finally, conducting data analysis and model validation using SPSS 22.0 and AMOS 24.0.

The main theories used in this study include the following:

Accident Causation Theory: Spanning from Heinrich’s Domino Theory to modern behavior-based safety models, this theoretical framework emphasizes unsafe behaviors as the critical link in accident chains [47], providing a foundation for analyzing risk factors in construction workers’ unsafe behaviors.

Social Exchange Theory: This theory elucidates the reciprocal relationship between organizational climate and individual behaviors, explaining how organizational support influences workers’ safety decisions through perceived fairness [48]. The organizational climate plays a pivotal role in this exchange process: organizations foster supportive environments, and employees reciprocate with heightened engagement and loyalty, thereby promoting safer and more productive behaviors. Aligning with this theory, organizational climate is identified as a key dimension influencing unsafe behaviors.

Social Cognitive Theory: Emphasizing individual agency, this theory systematically reveals the behavioral generation process, positing that individuals adapt their actions based on personal perceptions and environmental contexts. Bandura’s triadic reciprocal determinism [49]—highlighting dynamic interactions among personal factors (P), behaviors (B), and environmental influences (E)—serves as the basis for examining the mediating effects of anxiety psychology between environmental stressors and unsafe behaviors.

Two-Factor Theory of Motivation: By distinguishing hygiene factors (e.g., salary) from motivational factors (e.g., career development), this theory guides the formulation of intervention strategies to alleviate anxiety and reduce unsafe behaviors through job satisfaction optimization [50]. Drawing on Herzberg’s framework, this study positions job satisfaction as a critical dimension for mitigating unsafe behaviors and regulating anxiety psychology.

3.2. Questionnaire Design

The questionnaire targeted frontline construction workers and comprised three sections: the first section collected respondents’ demographic information, including gender, age, work experience, and education level; the second section assessed safety-related psychological states. The study employed the Self-Rating Anxiety Scale (SAS) to assess anxiety psychology, with scores categorized as follows: 0–50 points as normal, 50–59 points as mild anxiety, 60–69 points as moderate anxiety, and 70–100 points as severe anxiety; the third section quantified organizational climate, safety competency, job satisfaction, and unsafe behaviors using measurement scales—specifically, anxiety psychology (3 items), organizational climate (3 items), job satisfaction (5 items), safety competency (3 items), and unsafe behaviors (6 items), as shown in

Table 1. List of measurements.

Constructs	Label	Measurement Items
Anxiety Psychology	A1	Frequent abnormal heart rate fluctuations during work
	A2	Frequent elevated blood pressure during work
	A3	Frequent rapid breathing rates during work
Unsafe Behaviors	B1	Unsafe behaviors caused by operational loss of control at construction sites
	B2	Intentional unsafe behaviors due to resentment or emotional instability at construction sites
	B3	Unsafe behaviors stemming from lack of safety knowledge
	B4	Neglect of personal protective equipment (PPE) usage
	B5	Unsafe behaviors induced by external distractions
	B6	Unsafe behaviors attributable to hazardous construction environments
Organizational Climate	O1	Well-established safety protocols in the project department or team
	O2	Regular safety training conducted by the project department or team
	O3	Poor safety climate within the project department or team
Job Satisfaction	S1	Dissatisfaction with salary levels
	S2	Interpersonal conflicts with colleagues or supervisors
	S3	Lack of interest in work tasks
	S4	Operator dissatisfaction with on-site safety conditions
	S5	Operator dissatisfaction with leadership practices
Safety Competency	C1	Unsafe behaviors arising from inadequate job proficiency
	C2	Unsafe behaviors due to physical fatigue
	C3	Unsafe behaviors resulting from insufficient safety awareness

. All Measurement items employed a 5-point Likert scale (1 = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly Agree), with participants rating each item based on its congruence with their personal experiences and environmental contexts.

3.3. Data Collection

A pilot study with 168 valid questionnaires (from 200 distributed) was conducted prior to the formal survey to assess the reliability and validity of the initial scales. This sample size (N = 168) exceeded the recommended threshold of 10 times the number of observed variables, ensuring statistical rigor. The pilot data were exclusively used to refine ambiguous items and were excluded from the final analysis to prevent contamination. This approach aligns with established psychometric practices, where pre-testing on a sufficiently large subsample enhances scale robustness before full-scale deployment. The formal survey was conducted in November 2021, with frontline construction workers at the Xuzhou Garden Expo construction site serving as the core research sample. Data were collected utilizing both online platforms and offline paper-based questionnaires. A total of 300 questionnaires were distributed, of which 70% (210) were completed onsite via paper forms during safety meetings or rest periods, while 30% (90) were administered through a secure online platform. All participants provided informed consent both verbally and in writing, emphasizing principles of anonymity and voluntary participation. All data were stored in encrypted formats without collecting personally identifiable information, ensuring GDPR compliance. Of 300 questionnaires distributed, 280 were returned (93.33% response rate). After excluding incomplete, duplicate, and carelessly completed responses, 263 valid questionnaires were retained, representing a valid response rate of 87.66%.

The demographic characteristics of the sample are summarized in

Table 2. Descriptive statistical table of sample selection.

Category	Item	Frequency	Percentage
Gender	Male	186	70.7%
	Female	77	29.3%
Age	18–27	33	4.2%
	27–35	42	16.0%
	35–50	156	67.7%
	≥50	32	12.2%
Educational background	Junior high school or below	134	73.8%
	Higher school	54	20.5%
	College degree	43	4.9%
	Bachelor’s or above	32	0.8%
Experience	≤5	62	23.6%
	5–10	81	31.9%
	10–20	88	33.5%
	≥20	32	11.0%
Level of anxiety	Normal	35	13.3%
	Mild	96	36.5%
	Moderate	88	33.5%
	Severe	44	16.7%

.

Gender distribution: Males constituted 70.7% (n = 186) of respondents, while females accounted for 29.3% (n = 77), aligning with the male-dominated nature of construction sites.

Age structure: Over 50% of respondents were aged 35–50 years, reflecting significant workforce aging.

Educational background: 94.3% possessed junior or senior high school diplomas, consistent with the industry’s low educational attainment patterns.

Work experience: Most workers (>60%) had 5–20 years of experience, with fewer than 5% having less than 5 years.

Anxiety prevalence: 86.7% reported anxiety symptoms, underscoring the psychological strain induced by harsh working conditions.

The demographic profile closely mirrors industry-wide characteristics, confirming the sample’s generalizability and validity for analyzing anxiety psychology and unsafe behavior mechanisms.

3.4. Data Analysis Tools

This study utilized SPSS 22.0 and AMOS 24.0 for data analysis and model validation. SPSS 22.0 was employed to conduct reliability testing and KMO and Bartlett’s sphericity tests to ensure data quality. Correlation and regression analyses within SPSS 22.0 were performed to elucidate the association mechanisms between anxiety psychology (and its dimensions) and unsafe behaviors. AMOS 24.0 was applied to develop structural equation modeling (SEM) and perform variance testing to validate the theoretical model’s robustness. SEM, a multivariate analytical method, is widely employed in behavioral research due to its capacity to elucidate latent variable structures and quantify structural relationships among observed variables [51]. This methodology enables the systematic examination of complex causal pathways while statistically controlling measurement errors, thereby providing robust insights into the interplay between psychological constructs and behavioral outcomes.

4. Model Validation and Result Analysis

4.1. Questionnaire Reliability and Validity Testing

Given that the questionnaire data in this study reflect respondents’ attitudes or perceptions and focus on internal consistency among construction workers without repeated measurements, Cronbach’s α coefficient was employed to assess reliability. Validity analysis was conducted to evaluate the rationality and accuracy of the questionnaire’s structural design, with construct validity measured through exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) to examine relationships between questionnaire items and latent dimensions.

4.1.1. Exploratory Factor Analysis (EFA)

Based on the preliminary survey data, SPSS 22.0 was employed to calculate the Cronbach’s α coefficients and corrected item-total correlation (CITC) values for each measurement item. A measurement scale was considered reliable if its items met the criteria of Cronbach’s α > 0.7 and CITC > 0.5. In the pilot survey, exploratory factor analysis (EFA) was conducted to assess structural validity. First, the Kaiser–Meyer–Olkin (KMO) measure and Bartlett’s test of sphericity were performed to evaluate the suitability of the data for factor analysis. A KMO value closer to 1 indicates stronger suitability for factor analysis. Generally, the data are deemed appropriate for factor analysis if Bartlett’s test yields a significance level (Sig.) < 0.05 and the KMO value exceeds 0.7. Principal component analysis was then applied to extract factors with eigenvalues greater than 1 for each latent variable, as these factors were considered to significantly contribute to explaining data variance. All items associated with the latent variables were included in the factor extraction process, and only one factor with an eigenvalue > 1 was extracted for each variable, demonstrating satisfactory unidimensionality. The results of the exploratory factor analysis are presented in

Table 3. EFA and reliability testing results.

Constructs	Items	CITC	Cronbach’s α	KOM Value	Bartlett’s Test of Sphericity (Sig)	Factor Loadings	Eigenvalue	Cumulative Variance Explained
Unsafe Behaviors	B1	0.660	0.952	0.940	0.000	0.909	2.449	82.092%
	B2	0.712				0.888
	B3	0.823				0.921
	B4	0.722				0.822
	B5	0.684				0.831
	B6	0.745				0.920
Anxiety Psychology	A1	0.714	0.912	0.854	0.000	0.833	2.641	73.953%
	A2	0.851				0.891
	A3	0.654				0.942
Organizational Climate	O1	0.529	0.788	0.775	0.000	0.909	2.556	92.031%
	O2	0.634				0.921
	O3	0.772				0.941
Job Satisfaction	S1	0.582	0.884	0.712	0.000	0.898	2.779	80.892%
	S2	0.614				0.845
	S3	0.706				0.851
	S4	0.682				0.951
	S5	0.823				0.907
Safety Competency	C1	0.732	0.850	0.738	0.000	0.927	2.582	86.986%
	C2	0.638				0.918
	C3	0.709				0.901

. Following reliability and validity assessments of the pilot survey data, all variables were found to meet the required criteria and were deemed suitable for subsequent factor analysis.

4.1.2. Confirmatory Factor Analysis (CFA)

This study conducted confirmatory factor analysis (CFA) on the formal survey data using AMOS 24.0 software, with the results summarized in Table 3. The evaluation criteria included standardized factor loadings, composite reliability (CR), reliability coefficients, measurement errors, and average variance extracted (AVE). A measurement scale was deemed to exhibit satisfactory convergent validity only if the following thresholds were simultaneously met: standardized factor loadings > 0.5, CR > 0.6, and AVE > 0.5. As shown in

Table 4. CFA and reliability analysis results.

Constructs	Items	Std. FL	SMC	S.E	CR	AVE
Anxiety Psychology	A1	0.758	0.626	0.278
	A2	0.811	0.638	0.334	0.891	0.628
	A3	0.845	0.577	0.291
Organizational Climate	O1	0.732	0.612	0.264	0.842	0.667
	O2	0.797	0.633	0.336
	O3	0.767	0.512	0.285
Safety Competency	C1	0.801	0.533	0.364	0.869	0.634
	C2	0.713	0.523	0.322
	C3	0.725	0.565	0.318
Job Satisfaction	S1	0.811	0.561	0.285	0.823	0.629
	S2	0.762	0.613	0.281
	S3	0.765	0.541	0.326
	S4	0.773	0.546	0.261
	S5	0.783	0.576	0.253
Unsafe Behaviors	B1	0.854	0.711	0.283	0.878	0.548
	B2	0.821	0.562	0.301
	B3	0.712	0.651	0.313
	B4	0.756	0.638	0.345
	B5	0.743	0.522	0.292
	B6	0.698	0.579	0.340

, the standardized factor loadings for construction workers’ anxiety psychology and its dimensions (organizational climate, job satisfaction, safety competency) as well as unsafe behaviors all exceeded 0.5, with reliability coefficients > 0.5, CR values > 0.6, and AVE values > 0.5. These results confirm that the survey data for anxiety psychology, organizational climate, job satisfaction, safety competency, and unsafe behaviors demonstrate robust convergent validity and model fit, with all indices meeting the established criteria.

4.2. Correlation Analysis

Correlation analysis, a statistical method, quantifies and evaluates the existence and strength of associations between variables. To examine the linear relationship between construction workers’ anxiety psychology and unsafe behaviors—specifically, whether the correlation is positive or negative—this study employed SPSS 22.0 to analyze variable data and compute Pearson correlation coefficients (denoted as r). The results are presented in

Table 5. Correlation analysis between various dimensions of anxiety psychology and unsafe behavior of construction workers.

Variable	Item	Unsafe Behaviors
Anxiety Psychology	Pearson correlation	0.489 **
	Significance (two-tailed)	0.002
	Number of cases	263
Organizational Climate	Pearson correlation	−0.472 **
	Significance (two-tailed)	0.002
	Number of cases	263
Safety Competency	Pearson correlation	−0.442 **
	Significance (two-tailed)	0.003
	Number of cases	263
Job Satisfaction	Pearson correlation	−0.413 **
	Significance (two-tailed)	0.002
	Number of cases	263

** Significance level is p < 0.001.

. The Pearson correlation coefficient (r) measures the linear relationship between two continuous variables, ranging from −1 to 1. The absolute value of r indicates the strength of the correlation, while its sign reflects the direction. A value between 0 and 1 signifies a positive correlation (i.e., variables increase concurrently), whereas a value between −1 and 0 indicates a negative correlation (i.e., variables exhibit inverse trends).

As shown in Table 5, the Pearson correlation coefficient between anxiety psychology and unsafe behaviors is 0.498 (p < 0.01), confirming a significant positive correlation. This implies that fluctuations in anxiety psychology directly influence the likelihood of unsafe behaviors. Organizational climate (r = −0.472, p < 0.01), safety competency (r = −0.442, p < 0.01), and job satisfaction (r = −0.413, p < 0.01) all demonstrate significant negative correlations with unsafe behaviors. Among these variables, organizational climate exhibits the strongest negative association, suggesting that a harmonious organizational climate directly reduces the incidence of unsafe behaviors.

4.3. Analysis of Variance

Analysis of Variance (ANOVA) evaluates the normal distribution of collected data to determine the significant effects of factors on experimental indicators. This study employed one-way ANOVA to analyze the influence of demographic characteristics (gender, career age, educational background, work experience, and anxiety level) on anxiety psychology and unsafe behaviors. Results are presented in

Table 6. Analysis of anxiety and unsafe behavior of construction workers of demographic characteristics (N = 263).

Constructs	Variable	Ave				F	p
		A	B	C	D
Gender	Anxiety psychology	2.52	2.86	/	/	10.324	0.001
	Unsafe behaviors	3.22	3.01	/	/	2.151	0.201
Age	Anxiety psychology	2.43	2.76	2.88	2.65	11.438	0.005
	Unsafe behaviors	3.35	3.56	3.77	3.84	15.146	0.004
Educational background	Anxiety psychology	2.34	2.53	2.66	2.89	11.915	0.002
	Unsafe behaviors	3.73	3.63	3.55	3.34	15.355	0.003
Experience	Anxiety psychology	2.75	2.68	2.52	2.50	10.825	0.003
	Unsafe behaviors	3.44	3.13	3.08	3.21	2.615	0.201
Level of anxiety	Anxiety psychology	2.35	3.44	3.56	4.62	12.131	0.001
	Unsafe behaviors	2.14	3.33	3.54	4.81	2.615	0.001

A = Male; B = Female.

. A statistically significant difference between variables is confirmed when the F-statistic and p-value fall below the 0.05 significance threshold.

As shown in Table 6, gender exerted a significant effect on construction workers’ anxiety psychology (p < 0.05), with female workers exhibiting higher anxiety levels than their male counterparts in both daily life and work contexts. However, no significant gender-based differences were observed in unsafe behaviors (p > 0.05). Work experience significantly influenced anxiety psychology (p < 0.05) but showed no statistically significant association with unsafe behaviors (p > 0.05). Additionally, career age, educational background, and anxiety level demonstrated significant impacts on both anxiety psychology and unsafe behaviors (p < 0.05), indicating their critical roles in shaping psychological and behavioral outcomes.

4.4. Regression Analysis

To further clarify the relationship between anxiety psychology and unsafe behaviors, as well as the mediating roles of organizational climate, safety competency, and job satisfaction, multiple linear regression analysis was conducted to examine causal relationships among variables.

4.4.1. Main Effect Testing

Model 1 in

Table 7. Regression analysis of the main effect.

Dependent Variable	Dependent Variable: Unsafe Behaviors
	Model 1		Model 2
	β	t	β	t
Control variable
Gender	0.043	1.574	0.052	1.255
Age	0.073 **	0.532	0.077 **	0.487
Educational background	−0.136 *	−1.272	−0.115 *	−1.422
Experience	−0.075	−1.521	−0.018	−1.671
Level of anxiety	0.173 **	1.921	0.133 **	1.822
Independent variable
Anxiety psychology			0.679 ***	2.922
R2	0.385		0.601
R2 (Adjusted)	0.363		0.586
F	2.852 ***		82.630 ***

*** Significance level is p < 0.001. ** Significance level is p < 0.01. * Significance level is p < 0.05. All regression coefficients (β values) in the table represent standardized results.

represents the regression analysis of unsafe behaviors controlled for demographic variables (gender, career age, education level, work experience, and anxiety level). Model 2 introduces anxiety psychology as an independent variable while controlling for demographic factors to assess its direct effect on unsafe behaviors.

Model 1 demonstrated that career age (β = 0.073, p < 0.01), education background (β = −0.136, p < 0.05), and anxiety level (β = 0.173, p < 0.001) significantly predicted unsafe behaviors, while other demographic variables showed no statistical significance.

Model 2 revealed that anxiety psychology significantly predicted unsafe behaviors (β = 0.679, p < 0.001) after controlling for demographics. Career age exhibited a significant positive impact (β = 0.077, p < 0.01), whereas education level showed a negative association (β = −0.115, p < 0.05). The overall regression model achieved statistical significance (R² = 0.601, adjusted R² = 0.586, F = 82.63, p < 0.001), confirming the significant positive effect of anxiety psychology on unsafe behaviors. Thus, Hypothesis H1 was validated.

4.4.2. Mediation Effect Testing

Model 3 in

Table 8. Regression analysis of the mediating effect of job satisfaction.

Dependent Variable	Job Satisfaction		Unsafe Behaviors
	Model 3		Model 4		Model 5
	β	t	β	t	β	t
Control variable
Gender	0.041	1.031	0.051	1.034	0.049	0.049
Age	0.071 **	0.636	0.064 **	0.521	0.071 **	0.071 **
Educational background	−0.131 *	−1.221	−0.130 *	−1.451	−0.132 *	−0.132 *
Experience	−0.073	−1.523	−0.013	−1.634	−0.025	−0.025
Level of anxiety	−0.361 **	−1.618	−0.303 **	−1.728	−0.315 **	−1.468
Independent variable
Anxiety psychology	−0.389 **	−1.581	0.179 **	1.621
Mediating variable
Job satisfaction			−0.015	−0.195	−0.003 **	−0419
R2	0.448		0.641		0.623
R2 (Adjusted)	0.435		0.632		0.618
F	3.646 ***		4.369 ***		4.139 ***

*** Significance level is p < 0.001. ** Significance level is p < 0.01. * Significance level is p < 0.05.

examines the effect of anxiety psychology on job satisfaction after controlling for demographic variables (gender, career age, education level, work experience, and anxiety level). Model 4 introduces both anxiety psychology and job satisfaction (mediator) to assess their combined influence on unsafe behaviors. Model 5 evaluates the direct effect of job satisfaction (as the sole mediator) on unsafe behaviors while controlling for demographics.

Model 3 demonstrated a significant negative correlation between anxiety psychology and job satisfaction (β = −0.389, p < 0.01), validating Hypothesis H9 (“Job satisfaction is negatively correlated with anxiety psychology”).

Comparing Model 2 and Model 4, the regression coefficient of anxiety psychology on unsafe behaviors decreased significantly from β = 0.679 (p < 0.001) to β = 0.179 (p < 0.01), meeting the criteria for full mediation. This confirms that job satisfaction fully mediates the relationship between anxiety psychology and unsafe behaviors, supporting Hypothesis H10.

Model 5 revealed that job satisfaction alone significantly predicts unsafe behaviors (β = −0.003, p < 0.01) with a robust model fit (R² = 0.623, adjusted R² = 0.618, F = 4.139, p < 0.001), thereby validating Hypothesis H8 (“Job satisfaction is negatively correlated with unsafe behaviors”).

Model 6 in

Table 9. Regression analysis of the mediating effect of organizational climate.

Dependent Variable	Organizational Climate		Unsafe Behaviors
	Model 6		Model 7		Model 8
	β	t	β	t	β	t
Control variable
Gender	0.033	1.035	0.055	1.039	0.044	0.044
Age	0.073 **	0.631	0.061 **	0.522	0.073 **	0.072 **
Educational background	−0.133 *	−1.229	−0.135 *	−1.445	−0.136 *	−0.134 *
Experience	−0.069	−1.517	−0.019	−1.644	−0.021	−0.021
Level of anxiety	−0.359 **	−1.626	−0.316 **	−1.735	−0.310 **	−1.465
Independent variable
Anxiety psychology	−0.383 **	−1.595	0.071 **	1.629
Mediating Variable
Organizational climate			−0.013	−0.191	−0.002 **	−0412
R2	0.459		0.645		0.625
R2 (Adjusted)	0.446		0.639		0.618
F	3.655 ***		4.357 ***		4.263 ***

*** Significance level is p < 0.001. ** Significance level is p < 0.01. * Significance level is p < 0.05.

examines the effect of anxiety psychology on organizational climate after controlling for demographic variables (gender, career age, education level, work experience, and anxiety level). Model 7 introduces both anxiety psychology and organizational climate (mediator) to assess their combined influence on unsafe behaviors. Model 8 evaluates the direct effect of organizational climate (as the sole mediator) on unsafe behaviors while controlling for demographics.

Model 6 revealed a significant negative correlation between anxiety psychology and organizational climate (β = −0.383, p < 0.01), validating Hypothesis H3 (“Organizational climate is negatively correlated with anxiety psychology”).

Comparing Model 2 and Model 7, the regression coefficient of anxiety psychology on unsafe behaviors decreased substantially from β = 0.679 (p < 0.001) to β = 0.071 (p < 0.01), meeting the criteria for full mediation. This confirms that organizational climate fully mediates the anxiety psychology–unsafe behavior relationship, supporting Hypothesis H4.

Model 8 demonstrated that organizational climate alone significantly predicts unsafe behaviors (β = −0.002, t < 0.01) with strong model fit (R² = 0.625, adjusted R² = 0.618, F = 4.263, p < 0.001), thereby validating Hypothesis H2 (“Organizational climate is negatively correlated with unsafe behaviors”).

Model 9 in

Table 10. Regression analysis of the mediating effect of security capability.

Dependent Variable	Security Capability		Unsafe Behaviors
	Model 9		Model 10		Model 11
	β	t	β	t	β	t
Control variable
Gender	0.027	1.134	0.049	1.079	0.056	0.041
Age	0.064 **	0.622	0.059 **	0.563	0.061 **	0.063 **
Educational background	−0.153 *	−1.169	−0.121 *	−1.523	−0.126 *	−0.147 *
Experience	−0.045	−1.159	−0.059	−1.627	−0.033	−0.037
Level of anxiety	−0.321 **	−1.601	−0.367 **	−1.715	−0.356 **	−1.635
Independent variable
Anxiety psychology	−0.346 **	−1.655	0.105 *	1.563
Mediating Variable
Security capability			−0.011	−0.167	−0.003 **	−0.537
R2	0.562		0.671		0.664
R2 (Adjusted)	0.489		0.615		0.616
F	3.378 ***		4.251 ***		4.194 ***

*** Significance level is p < 0.001. ** Significance level is p < 0.01. * Significance level is p < 0.05.

examines the effect of anxiety psychology on safety competency after controlling for demographic variables (gender, career age, education level, work experience, and anxiety level). Model 10 introduces both anxiety psychology and safety competency (mediator) to assess their combined influence on unsafe behaviors. Model 11 evaluates the direct effect of safety competency (as the sole mediator) on unsafe behaviors while controlling for demographics.

Model 9 revealed a significant negative correlation between anxiety psychology and safety competency (β = −0.346, p < 0.01), validating Hypothesis H6 (“Safety competency is negatively correlated with anxiety psychology”).

Comparing Model 2 and Model 10, the regression coefficient of anxiety psychology on unsafe behaviors decreased markedly from β = 0.679 (p < 0.001) to β = 0.105 (p < 0.05), meeting the criteria for full mediation. This confirms that safety competency fully mediates the anxiety psychology–unsafe behavior relationship, supporting Hypothesis H7.

Model 11 demonstrated that safety competency alone significantly predicts unsafe behaviors (β = −0.003, p < 0.01) with robust model fit (R² = 0.664, adjusted R² = 0.616, F = 4.194, p < 0.001), thereby validating Hypothesis H5 (“Safety competency is negatively correlated with unsafe behaviors”).

4.5. Structural Equation Modeling (SEM)

Building on satisfactory reliability and validity from pilot and formal surveys, this study employed structural equation modeling (SEM) via AMOS 24.0 to validate the hypothesized relationships. The initial SEM (Figure 3) illustrates anxiety psychology as a latent variable (ellipse), unsafe behaviors as an observed variable (rectangle), and residuals (circles) with measurement errors labeled e1–e17. Path coefficients between anxiety psychology and unsafe behaviors were positive, while those linking organizational climate, safety competency, job satisfaction, and anxiety psychology to unsafe behaviors were negative, aligning with theoretical hypotheses and confirming the model’s preliminary validity.

The model was validated using AMOS 24.0 by assessing three categories of fit indices: absolute fit, incremental fit, and parsimonious fit, with results detailed in

Table 11. Fitting of initial data.

	S.E	C.R	p	Normalized Path Efficient
Organizational climate → O1	0.098	8.451	***	0.811
Organizational climate → O2	0.093	8.484	***	0.743
Organizational climate → O3	0.091	7.981	***	0.783
Security capability → C1	0.083	7.684	***	0.714
Security capability → C2	0.078	7.757	***	0.818
Security capability → C3	0.074	9.013	***	0.732
Job satisfaction → S1	0.071	9.145	***	0.779
Job satisfaction → S2	0.066	9.367	***	0.755
Job satisfaction → S3	0.062	9.313	***	0.782
Job satisfaction → S4	0.052	9.011	***	0.778
Job satisfaction → S5	0.051	9.112	***	0.821
Anxiety psychology → A1	0.058	9.142	***	0.771
Anxiety psychology → A2	0.042	8.883	***	0.742
Anxiety psychology → A3	0.039	8.734	***	0.821
Unsafe behaviors → B1	0.045	8.678	***	0.777
Unsafe behaviors → B2	0.041	8.618	***	0.735
Unsafe behaviors → B3	0.038	7.457	***	0.803
Unsafe behaviors → B4	0.037	7.562	***	0.796
Unsafe behaviors → B5	0.036	8.542	***	0.838
Unsafe behaviors → B6	0.049	8.533	***	0.757

*** Significance level is p < 0.001.

,

Table 12. Analysis of the overall fit index of the model.

Fitting Index		Values	Recommended Values	Result
Absolute Fit	X2/df	2.627	<3	Acceptable
	RMSEA	0.041	<0.08	Acceptable
	GFI	0.834	>0.9	Acceptable
	RMR	0.043	<0.05	Acceptable
Incremental Fit	IFI	0.925	>0.9	Acceptable
	CFI	0.957	>0.9	Acceptable
	TLI	0.968	>0.9	Acceptable
	NFI	0.954	>0.9	Acceptable
Parsimonious Fit	PGFI	0.913	>0.9	Acceptable

, and

Table 13. Action path coefficient.

	S.E.	C.R	p	Normalized Path Coefficient
Organizational climate → Anxiety psychology	0.063	−4.423	***	−0.358
Security capability → Anxiety psychology	0.071	−3.270	***	−0.142
Job satisfaction → Anxiety psychology	0.068	−4.713	*	−0.279
Organizational climate → Unsafe behaviors	0.061	−3.161	**	−0.216
Security capability → Unsafe behaviors	0.071	−3.586	***	−0.197
Job satisfaction → Unsafe behaviors	0.079	−3.178	*	−0.178
Anxiety psychology → Unsafe behaviors	0.065	3.286	***	0.336

*** Significance level is p < 0.001. ** Significance level is p < 0.01. * Significance level is p < 0.05.

, respectively. As presented in Table 11, all standardized estimates of observed variables ranged between 0.50 and 0.95, with error variances (S.E) spanning 0.021 to 0.098 and no negative variances detected, indicating satisfactory basic model fit. Table 12 demonstrates that all global fit indices (e.g., χ²/df, RMSEA, CFI) met recommended thresholds, confirming excellent overall model fit.

As shown in Table 13, all path coefficients achieved statistical significance (p < 0.05), with standard errors (S.E.) falling within reasonable ranges (no negative or extreme values detected), thereby satisfying model validity criteria.

The hypothesis validation results for the relationship between anxiety psychology and unsafe behaviors are as follows:

H1: Anxiety psychology → Unsafe behaviors (standardized coefficient = 0.336, S.E. = 0.065, C.R. = 3.286, p < 0.001)

The path achieved statistical significance (p < 0.001), confirming Hypothesis H1.

The hypothesis validation results for the mediating variables (organizational climate, safety competency, job satisfaction) in relation to anxiety psychology are as follows:

H3: Organizational climate → Anxiety psychology (standardized coefficient = −0.358, S.E. = 0.063, C.R. = −4.423, p < 0.001)

H6: Safety competency → Anxiety psychology (standardized coefficient = −0.142, S.E. = 0.071, C.R. = −3.270, p < 0.001)

H9: Job satisfaction → Anxiety psychology (standardized coefficient = −0.279, S.E. = 0.068, C.R. = −4.713, p < 0.05)

All paths were statistically significant (p < 0.05), validating Hypotheses H3, H6, and H9, with organizational climate and job satisfaction exerting pronounced effects.

The hypothesis validation results for the mediating variables (organizational climate, safety competency, job satisfaction) in relation to unsafe behaviors are as follows:

H2: Organizational climate → Unsafe behaviors (standardized coefficient = −0.216, S.E. = 0.061, C.R. = −3.161, p < 0.01)

H5: Safety competency → Unsafe behaviors (standardized coefficient = −0.197, S.E. = 0.071, C.R. = −3.586, p < 0.001)

H8: Job satisfaction → Unsafe behaviors (standardized coefficient = −0.178, S.E. = 0.079, C.R. = −3.178, p < 0.05)

All paths were statistically significant (p < 0.05), validating Hypotheses H2, H5, and H8. Organizational climate demonstrated the strongest negative impact on unsafe behaviors.

5. Discussion

5.1. Impact Mechanism of Anxiety Psychology on Unsafe Behaviors

The impact mechanism underlying anxiety psychology on unsafe behaviors among construction workers, as identified through the developed theoretical model, proposed relational hypotheses, and SEM-based theoretical model testing and results analysis (Figure 4), is systematically summarized and discussed below.

• Negative Correlation Between Anxiety Psychology and Organizational Climate

A significant negative correlation exists between construction workers’ anxiety psychology and organizational climate (path coefficient = −0.358). A poor organizational climate exacerbates collective anxiety, as workers in chaotic environments experience heightened insecurity, fostering anxiety contagion and even organization-wide anxiety. Conversely, robust safety protocols and positive climates stabilize workers’ anxiety levels, enhance engagement, and reduce accident probabilities.

• Negative Correlation Between Organizational Climate and Unsafe Behaviors

Organizational climate demonstrates a significant negative association with unsafe behaviors (path coefficient = −0.216). Regular safety training and institutionalized practices minimize risks, improve efficiency, and reduce unsafe acts. Conversely, poor climates amplify emotional volatility, weaken safety perceptions, and increase unsafe behavior likelihood.

• Negative Correlation Between Anxiety Psychology and Safety Competency

Safety competency significantly mitigates anxiety psychology (path coefficient = −0.142). Workers with deficient safety knowledge and physical fitness struggle to manage stressors, whereas those with strong competencies maintain adaptive anxiety thresholds during emergencies.

• Negative Correlation Between Safety Competency and Unsafe Behaviors

Higher safety competency (path coefficient = −0.197)—encompassing knowledge, skills, and physical readiness—reduces unsafe acts by enabling effective hazard responses. Conversely, low competency leads to mismatched physical–cognitive reactions during crises, escalating accident risks.

• Negative Correlation Between Anxiety Psychology and Job Satisfaction

Job satisfaction inversely predicts anxiety psychology (path coefficient = −0.279). Declines in hygiene factors (e.g., wages) and motivational factors (e.g., leadership quality) heighten anxiety, whereas fair compensation, positive relationships, and supportive leadership stabilize workers’ psychological states.

• Negative Correlation Between Job Satisfaction and Unsafe Behaviors

Low job satisfaction (path coefficient = −0.178) correlates with increased unsafe behaviors, as disengaged workers exhibit negligence and negative emotions. Prolonged dissatisfaction perpetuates risk-prone behaviors, elevating accident probabilities.

• Positive Correlation Between Anxiety Psychology and Unsafe Behaviors

Anxiety psychology significantly predicts unsafe behaviors (path coefficient = 0.336). Severe anxiety manifests as physiological impairments (e.g., dyspnea, hyperhidrosis, muscle tension, intermittent dizziness), impairing hazard judgment and directly escalating safety incidents.

• Significant Differential Effects of Demographic Variables on Anxiety Psychology and Unsafe Behaviors

Among the five demographic variables, construction workers’ age, educational background, and anxiety levels demonstrate statistically significant impacts on unsafe behaviors, whereas other demographic variables (gender and work experience) exhibit no significant effects: (1) gender demonstrates no significant disparity in influencing unsafe behaviors. The absence of substantial differences in work engagement between male and female workers results in negligible gender-based variations in unsafe behavioral outcomes; (2) age significantly affects anxiety psychology and unsafe behaviors (path coefficient = 0.077, p < 0.01). Older workers, being more sensitive to life stressors, are prone to heightened anxiety, which amplifies risk-prone behavioral tendencies; (3) educational background significantly influences anxiety psychology and unsafe behaviors (path coefficient = −0.115, p < 0.05). Highly educated workers, with superior comprehension and adaptability, exhibit enhanced safety competencies; however, their increased exposure to socioeconomic anxieties in suboptimal work environments (e.g., poor safety conditions or low occupational status) exacerbates anxiety-driven unsafe actions; (4) work experience shows no statistically significant correlation with unsafe behaviors; (5) anxiety levels exhibit a strong positive correlation with both anxiety psychology and unsafe behaviors (path coefficient = 0.133, p < 0.01). Escalating anxiety directly intensifies physiological stress responses (e.g., elevated cortisol levels) and increases the frequency of unsafe practices.

5.2. Preventive Measures and Suggestions

This study demonstrates that maintaining construction workers’ anxiety psychology within mild thresholds effectively enhances productivity and safety performance. The triadic dimensions of organizational climate, safety competency, and job satisfaction exert significant mediating effects across distinct pathways linking anxiety psychology to unsafe behaviors. Grounded in the triadic reciprocal determinism of social cognitive theory—which emphasizes the dynamic interplay between individual and environmental factors—this study proposes integrative interventions targeting individual-level regulation (enhancing safety competency, ensuring job satisfaction) and environmental-level optimization (strengthening organizational climate). These measures collectively mitigate anxiety psychology and reduce unsafe behaviors, with the comprehensive improvement framework systematically illustrated in Figure 5.

5.2.1. External Environmental Control

External environmental controls reconcile conflicts between individual predispositions and organizational rules by providing resource guarantees and policy support for organizational climate development through measures including fostering a robust safety culture, establishing comprehensive safety protocols, conducting regular safety training, introducing external psychological counseling institutions, and strengthening governmental oversight. Specifically, cultivating a sustainable organizational climate necessitates the following strategies: Firstly, leadership-driven safety culture should be prioritized to establish a supportive safety atmosphere [52], requiring managers to lead by example in adhering to safety regulations and demonstrating accountability in accident resolution, thereby reducing workers’ anxiety through behavioral exemplification. Secondly, a dual-track regulatory system must be implemented—rigidly constraining behaviors via standardized safety accountability mechanisms (e.g., random inspections, performance evaluations) while designing inclusive policies (e.g., tiered and visualized safety education programs) to accommodate workers’ cognitive diversity, alleviating anxiety caused by skill gaps. Concurrently, assessment mechanisms should be integrated to form closed-loop management, transitioning workers from passive compliance to proactive internalization of safety knowledge, thereby enhancing risk anticipation capabilities and mitigating psychological burdens arising from ambiguous safety rules. Thirdly, external psychological counseling institutions should be utilized to periodically assess workers’ anxiety levels, trace root causes (e.g., familial stressors, skill deficiencies), and synchronize feedback to enterprises for adjusting management strategies (e.g., targeted skill training, family support policies), achieving closed-loop coordination between individualized psychological intervention and organizational climate optimization. Finally, governmental supervision should reinforce external incentives by incorporating mental health management into administrative regulations, compelling enterprises to prioritize psychological safety and organizational climate development through legally mandated accountability. These measures collectively contribute to the formation of a sustainable, safety-oriented organizational climate.

5.2.2. Internal Self-Regulation

Individual internal regulation plays a foundational role in unsafe behavior prevention, operating via a dual-mechanism synergy of safety competency enhancement and job satisfaction assurance. For safety competency development, a comprehensive intervention system should be implemented, encompassing psychological resilience training, technical skill upgrading, safety cognition deepening, and physical fitness optimization. Psychologically, workers are encouraged to adopt evidence-based anxiety mitigation strategies such as breathing techniques, exercise-based stress reduction, and social sharing, while organizations should provide recreational outlets to facilitate emotional catharsis and internalize safety values. Regarding job satisfaction, scientifically restructuring compensation systems is critical, requiring government-led mechanisms for standardized payment cycles, transparent disbursement methods, and legally binding safeguards to eliminate economic anxiety triggers. Concurrently, project teams must institute conflict early-warning systems and rapid mediation protocols—including regular consultations and managerial interventions—to resolve interpersonal tensions promptly, mitigating cumulative anxiety’s detrimental impacts on operational safety in this transient, team-based industry.

5.2.3. Direct Anxiety Intervention

The intervention of anxiety plays a pivotal role in mitigating construction workers’ unsafe behaviors, centering on the establishment of a dual intervention mechanism integrating cognitive restructuring and physiological symptom management. Empirical studies confirm that anxiety in construction workers predominantly stems from catastrophic interpretations of negative emotions and overgeneralization of behavioral consequences, where cognitive behavioral therapy (CBT) demonstrates significant efficacy. Interventions should guide workers to accept anxiety states rather than engage in counterproductive suppression, utilize distraction techniques (e.g., leave periods, skill training) to reduce self-focused attention, foster positive experiences through interest-driven tasks, and establish a “normalization of anxiety” cognitive framework to gradually attenuate the reinforcement of anxiety.

At the physiological level, integrating symptom recognition and self-regulation techniques into pre-task safety briefings is essential, creating rapid-response protocols for symptom–behavior linkages. Specific strategies include breathing regulation training (e.g., diaphragmatic breathing), structured exercise interventions (e.g., jogging, swimming), and attentional diversion tactics, complemented by sleep rhythm optimization and dietary adjustments to achieve psychophysiological synergy. This integrated model of cognitive–behavioral intervention enhances psychological resilience by altering workers’ attribution styles toward anxiety while bolstering emotional self-regulation through physiological techniques, thereby concurrently reducing anxiety levels and blocking its translation into unsafe behaviors. Such an approach provides actionable pathways for mental health management on construction sites.

6. Conclusions

This study constructed a theoretical model to investigate the mechanisms through which anxiety psychology influences construction workers’ unsafe behaviors, incorporating demographic characteristics (gender, age, educational background, work experience, anxiety level) as moderating variables, anxiety psychology as the independent variable, unsafe behaviors as the dependent variable, and organizational climate, safety competency, and job satisfaction as mediating variables. The findings reveal a significant positive correlation between anxiety psychology and unsafe behaviors, with moderate anxiety levels enhancing organizational productivity and safety performance. Conversely, organizational climate, safety competency, and job satisfaction exhibit significant negative correlations with unsafe behaviors, indicating that higher levels of these mediators reduce both anxiety and unsafe acts. Gender and education level significantly influence anxiety states: female workers demonstrate heightened anxiety in hazardous environments due to emotional sensitivity, while highly educated workers exhibit lower unsafe behavior tendencies owing to stronger safety cognition. While this study focuses on construction workers, the anxiety–behavior influence mechanism may extend to other high-risk occupations, such as frontline workers in coal mining, petrochemical, power, and manufacturing industries, pending further empirical validation across these sectors.

The research contributions encompass three key aspects: (1) the development of a culturally adapted measurement scale for assessing construction workers’ anxiety psychology and unsafe behaviors within China’s construction context, refined through pilot testing to enhance ecological validity, respondent comprehension, and data reliability; (2) the proposal and validation of an integrated theoretical model elucidating the mechanisms through which individual, psychological, and environmental factors collectively influence unsafe behaviors, employing correlational, regression, ANOVA, and structural equation modeling (SEM) analyses; (3) the formulation of sociocognitive intervention strategies grounded in Bandura’s triadic reciprocal determinism, which integrates environmental regulation (e.g., safety protocols), individual adaptation (e.g., competency training), and dynamic anxiety management to mitigate unsafe behaviors, thereby advancing innovative approaches for construction safety governance.

This study acknowledges limitations. Firstly, the limited representativeness of the sample, despite efforts to ensure diversity by targeting multiple job categories (e.g., bricklayers, electricians, and mechanical workers) at the Xuzhou Garden Expo construction site, may constrain generalizability due to the convenience sampling method employed. Future research should adopt stratified random sampling across regions and project types to enhance external validity. Secondly, the reliance on self-rated scales for assessing anxiety psychology may introduce subjective bias; integrating physiological indicators (e.g., cortisol levels) could provide objective validation in subsequent studies. Additionally, the lack of dynamic analysis on anxiety fluctuations across distinct construction phases (e.g., high-altitude tasks vs. equipment operations) limits scenario-specific insights. Future investigations should incorporate context-specific temporal tracking to deepen the understanding of anxiety–behavior interactions in varying task environments.