The Cross-Cutting Effects of Age Expectation and Safety Value on Construction Worker Safety Behavior: A Multidimensional Analysis

Abstract

This research investigates how age expectation (AE) influences construction worker safety behavior (CWSB) by utilizing self-validation theory (SVT). Using structural equation modeling (SEM) and data from 391 Chinese construction workers, we found that components of AE, such as physical health expectation (PHE), mental health expectation (MHE), and cognitive function expectation (CFE), positively impact CWSB. Safety value (SV), including safety risk perception (SRP) and safety attitude (SA), mediates these effects. The study provides theoretical insights and practical guidance for enhancing CWSB through interventions on AE, supporting sustainable growth and workforce quality in the construction industry.

1. Introduction

The construction industry is a solid pillar of the global economy. The global construction market reached a value of approximately USD 13.57 trillion in 2023 [1,2]. It is projected to grow at a compound annual growth rate (CAGR) of 6.5% from 2024 to 2032, reaching an estimated value of USD 23.92 trillion by 2032 [3]. This highlights the construction industry’s dynamism and growth potential within the global economy, underscoring its role as a significant driver of global economic expansion. However, in contrast to its economic contribution, the construction industry’s safety record has been less than stellar, with accident and fatality rates much higher than those of other industries. According to the International Labor Organization (ILO), at least 60,000 construction workers die each year due to safety accidents worldwide [4,5]. According to the U.S. Bureau of Labor Statistics (BLS), 1008 U.S. construction workers were killed in occupational safety accidents in 2021, accounting for approximately 21% of fatal occupational safety accidents in all industries [6]. China’s construction industry also showed a double-high trend in accident and fatality rates, with 734 safety accidents occurring in the construction industry in 2023, resulting in the deaths of 840 workers [7]. After a statistical investigation of numerous safety accidents, it was found that human error was a significant cause of these incidents. The statistical data on safety accidents in the construction industry in both China and the United States showed that approximately 80–90% of the causes of accidents were closely related to human behavior [8]. Therefore, the study of CWSB within the construction industry has been confronted with a pressing concern.

The global demographic landscape is undergoing a significant transformation characterized by increasing life expectancy and declining birth rates [9]. This phenomenon has led to the progressive aging of the global population, exposing the construction industry to the obvious problems associated with an aging workforce. The aging process poses a multifaceted challenge for construction workers, manifesting in various ways, including reduced physical strength, diminished hearing and vision capabilities, impaired breathing functions, and extended recovery times [10]. These physical and cognitive limitations significantly impact the ability of construction workers to engage in physically demanding labor while upholding safety standards [11]. Therefore the concept of active aging has emerged as a pivotal approach aimed at empowering construction workers to maintain and enhance CWSB as they grow older [12]. It is imperative for managers to implement appropriate measures that cater to the evolving needs of aging construction workers [10,13]. This proactive approach not only enhances the efficiency of the workforce but also fosters improved safety practices. While numerous studies have endeavored to identify factors linked to CWSB within the construction industry, the majority of these investigations have concentrated primarily on physical safety risks [14,15,16]. Moreover, although some attention has been directed toward addressing the specific challenges and requirements of aging construction workers [17], there remains an evident gap in the literature concerning how construction workers’ AE influences their behaviors. Consequently, one of the objectives of this study is to explore the AE that construction workers hold regarding their own aging process and to examine how these expectations exert an influence on CWSB. Understanding this mechanism could lead to more effective strategies for mitigating unsafe behaviors and enhancing overall safety within the construction industry.

AE encompasses anticipations regarding the evolution of physical health, mental health, and cognitive function in the aging process [18]. These expectations exert a demonstrable influence on individual behaviors [19]. Extant literature has robustly established a positive correlation between self-perceptions of aging and the adoption of multiple preventive health behaviors, a relationship that persists even subsequent to adjusting for significant covariates [20]. Levy’s stereotype embodiment theory posits that the aging trajectory is inextricably intertwined with societal norms and biomedical paradigms, culminating in the formation of pervasive negative stereotypes associated with aging, characterized by frailty, dysfunction, and illness [21]. As individuals progress through the aging continuum, there is a propensity, whether conscious or subliminal, to align behaviors with these ingrained stereotypical expectations, thereby engendering a self-fulfilling prophecy. Given this theoretical backdrop, it is both logical and pertinent to hypothesize that AE may exert an influence on CWSB.

In exploring the influence of AE on CWSB, the formation and transmission of SV are also involved. SV is pivotal to the organizational safety climate, encompassing workers’ attitudes towards and beliefs in occupational safety and health. SV can manifest through responses to stimuli that shape mental states and attitudes, thereby fostering protective behaviors. Alternatively, SV may emerge from employing technical and scientific strategies informed by knowledge and engagement, aimed at mitigating risks associated with human performance in workplace settings [22]. Furthermore, psychological factors—including construction workers’ SRP and SA—play a critical role in shaping the emphasis they place on their own safety. These factors are subsumed under individual safety values [21]. Therefore, another objective of this study is to explore how AE influences SV and how these values affect CWSB. Understanding these influences facilitates the development of targeted interventions and policies to enhance workplace safety and reduce accident rates, ultimately benefiting the entire construction industry.

2. Literature Review and Hypotheses

2.1. Theoretical Basis

The objective of this research was to examine the interactions among AE, SV, and CWSB. Employing SVT [23] as the conceptual backbone, this work substantiated the link between AE and CWSB. SVT posits that individuals enhance their control and predictive capabilities over their environments by seeking or generating feedback that aligns with their self-concept [24], thereby consolidating their self-verification [25] through the assimilation of external information. Guided by the “cognition–action” rationale intrinsic to self-concept, it is postulated that construction workers, upon evaluating both objective conditions and subjective perceptions of CWSB, endeavor to reinforce constructs that align with their assessments, thereby boosting their control and predictive capacities concerning their external surroundings [26]. Within the SVT framework, this cognition–action linkage serves as a key proximal determinant in modeling CWSB, where the cognitive component encompasses AE and SV and the action component pertains to CWSB. SVT proposes five mechanisms through which the self-concept is maintained or reinforced, subsequently shaping behavior. These mechanisms include self-cognition, expectations, information selection, information interpretation, and memory bias. Based on SVT and its underlying mechanisms, we have developed a conceptual model and tested the effectiveness of AE in enhancing SV and CWSB.

2.2. AE and CWSB

In the context of SVT and its fundamental mechanisms, which encompass self-cognition, anticipatory processes, information acquisition, information interpretation, and memory tendency, AE plays a pivotal role within the cognitive facet of the self-verification logic chain. AE pertains to individuals’ expectations regarding aging, encompassing physical health, mental health, and cognitive function. Within the framework of SVT mechanisms, it becomes evident that positive AE reinforces or sustains the perception of safety within one’s self-concept. Conversely, when individuals encounter negative AE, it diminishes the sense of safety, subsequently leading to alterations in behavior.

2.2.1. PHE and CWSB

PHE, as conceptualized in the literature, pertains to anticipations concerning aging in relation to physical health [18]. Within the medical sphere, these expectations about aging significantly influence individuals’ health and well-being in their later years [27]. It is observed that individuals who perceive aging as a process encompassing potentialities, as opposed to those who view it broadly as a pathway to disability, are less prone to experience pronounced declines in physical, physiological, and social functioning [28]. Individuals harboring positivity are adept at dynamically optimizing the allocation of their finite cognitive resources, which, in turn, enhances their working and long-term memory performance [29]. Conversely, individuals with negative perceptions of their PHE tend to lower their life expectancy as they age, initiating a self-fulfilling cycle that exacerbates their physical and social health, as well as their behavior [30]. This decline in physical functionality [31] is frequently accompanied by a spectrum of chronic ailments such as hypertension, infections, digestive disorders, visual and auditory impairments, bone fractures, and frequent urination [32]. Applying this concept to the context of the construction industry, it can be posited that construction workers with positive PHE are more likely to effectively plan their tasks, adhere to safety protocols, and accumulate experiential knowledge. In contrast, workers with negative PHE may exhibit diminished physical capabilities and an elevated disease risk, factors that could adversely influence CWSB. Hence, this research proposes the following hypothesis:

Hypothesis a1 (Ha1):

PHE exerts a positive and significant impact on CWSB.

2.2.2. MHE and CWSB

MHE is defined as expectations of aging in terms of mental health [18]. The Mental Health Expectations Study, conducted among U.S. veterans, highlighted that those with positive MHE exhibited significantly reduced levels of psychological stress and a lower incidence of mental health issues, such as suicidal ideation, anxiety, and post-traumatic stress disorder (PTSD) [33,34]. This is in stark contrast to veterans harboring negative mental health expectations, who demonstrated a higher propensity for such conditions [35]. These findings underscore the critical role of MHE in shaping veterans’ overall mental well-being and functional capacity. Applying these insights to the construction industry, it is evident that construction workers with positive MHE are at a substantially lower risk of mental health disorders compared to their counterparts with negative expectations. This positive outlook equips them to more effectively manage work-related stress, engage actively in safety training, and adopt safe working practices. Based on these observations, the following hypothesis is proposed for investigation in this research:

Hypothesis a2 (Ha2):

MHE has a positive and significant effect on CWSB.

2.2.3. CFE and CWSB

CFE pertains to anticipations regarding aging and its impact on cognitive abilities. [18]. Within the medical domain, this expectation is intricately linked to cognitive capabilities, which are pivotal for myriad daily life activities and constitute a crucial aspect in evaluating life quality [36]. Individuals with positive CFE are shown to have a diminished risk of cognitive deterioration. This includes a slowdown in processing speed [37], memory impairment [38], and a heightened risk of developing Alzheimer’s disease [39]. Furthermore, such individuals typically exhibit enhanced emotional assessment capabilities, fostering an active and engaged lifestyle [39]. Conversely, negative CFE is correlated with the opposite effects. Applying these concepts to the construction industry, construction workers with a positive CFE are likely to possess robust cognitive abilities, enabling them to swiftly and effectively respond to emergencies. Their retention of accumulated work experience contributes to safer work practices. In contrast, workers with negative CFE tend to exhibit lower cognitive abilities and a higher probability of engaging in unsafe behaviors. Therefore, this research proposes the following hypothesis:

Hypothesis a3 (Ha3):

CFE has a positive and significant impact on CWSB.

2.3. SV and CWSB

Under the framework of SVT, the interplay between SV and CWSB is conceptualized as a dynamic and interactive process. This process underscores the profound influence of individual values on behavioral manifestations. SVT’s core mechanisms include self-cognition, anticipatory processes, memory tendency, and the processes of information acquisition and interpretation. SV, ingrained as a foundational belief system, shapes an individual’s perception and expectations in contexts related to safety. Higher levels of SV are correlated with an increased likelihood of anticipating potential negative outcomes from unsafe behaviors, leading to more cautious decision-making in behavioral choices. Individuals with strong SV are not only convinced of the importance of adhering to safety protocols but also possess a higher belief in their ability to effectively engage in CWSB. This belief fosters a proactive stance towards potential security threats, as evidenced by initiatives such as seeking safety training or enhancing workplace conditions to minimize risk exposures.

2.3.1. SRP and CWSB

SRP is widely acknowledged as a foundational element in the formulation of safety risk management strategies [40]. This perception encompasses the ability of individuals, groups, or organizations to discern varying degrees of risk, an ability influenced by their beliefs, attitudes, judgments, and emotional responses to natural, technological, or social risks and hazards [41]. The ability to accurately perceive potential safety risks is vital, as it enables the workers to implement appropriate measures to preclude possible accidents [42,43]. In contrast, a misperception or underestimation of safety risks may result in unsuitable safety decisions, subsequently increasing the likelihood of workplace injuries [44]. Drawing from these insights, it is plausible to conjecture that the SRP of construction workers exerts a considerable positive influence on CWSB. Based on this rationale, the following hypothesis is proposed:

Hypothesis b1 (Hb1):

SRP has a positive and significant effect on CWSB.

2.3.2. SA and CWSB

SA encompasses the spectrum of workers’ feelings, predispositions, and behavioral intentions concerning workplace safety [45]. This attribute reflects an individual’s inclination to respond to risk situations and safety norms within an organizational setting, either positively or negatively [46]. In the construction industry, there is a close link between SA and CWSB. Mearns and Flin [47] emphasized that workers who prioritize safety and value adherence to safety protocols are more likely to engage in behaviors that minimize risk and prevent accidents. Consequently, workers with strong SA are more diligent in following safety procedures and using protective equipment. In contrast, workers with weak SA tend to neglect safety hazards, leading to a higher likelihood of accidents. This leads to the formulation of the following hypothesis:

Hypothesis b2 (Hb2):

SA exerts a positive and significant influence on CWSB.

2.4. AE and SV

In the context of SVT, the nexus between AE and SV is conceptualized as a multifaceted process driven by the interplay between individual self-perception and behavior feedback mechanisms. This dynamic entails an individual’s cognizance of their own aging trajectory and anticipated capabilities, and critically, how these perceptions shape their SV and consequent safety-related behaviors. Primarily, AE—constituting an individual’s forecast of their future proficiencies and overall condition—exerts a profound influence on SV. This influence manifests when individuals, anticipating a decrement in their physical and cognitive faculties as they age, accord augmented importance to safety measures aimed at mitigating potential workplace accidents. Such anticipatory outlooks heighten the individual’s acuity regarding safety concerns, effectively elevating the precedence of safety within their personal value hierarchy. Secondarily, within the framework of SVT, the construct of AE is posited to have a direct bearing on an individual’s SV. This influence is particularly pronounced when individuals perceive a waning in their capacity to efficaciously implement safety strategies with advancing age, potentially engendering a heightened sense of vulnerability and diminished control in the face of safety hazards. This altered perception may attenuate their motivation to comply with established safety protocols and engage in proactive CWSB, thereby adversely impacting their SV.

2.4.1. PHE and SV

A review by the Department of Trade and Industry [48] has provided evidence that physical or mental decline associated with normal aging rarely affects performance in most jobs until the age of 70, except for jobs requiring quick reflexes or physical strength [49]. However, the possible increase in work-related illnesses and accidents among workers over 60 years of age remains a concern. Topics related to workplace accidents include mobility, strength, flexibility, balance, and sensory loss (including hearing and vision) [50]. When workers’ physical conditions change, information processing slows down, and reaction times lengthen [51]. These changes can make them less likely to correctly identify potential hazards in the workplace and less likely to respond positively to workplace safety standards. So, construction workers with positive PHE are likely to increase their safety awareness and thus improve their SV, whereas construction workers with negative PHE are likely to ignore safety protocols and be incapable of recognizing risks in the workplace in a timely manner [52]. Therefore, it can be inferred that PHE has positive significance for SRP and SA. And the following hypotheses are made:

Hypothesis c1 (Hc1):

PHE has positive significance for SRP.

Hypothesis c2 (Hc2):

PHE has positive significance for SA.

2.4.2. MHE and SV

Mental health problems are a significant source of disability at work [53]. They will affect an individual’s judgment of the surrounding environment, and the ability to perceive the occurrence of risks will also change [54]. During the aging process, construction workers inevitably face some mental health issues that can affect their ability to perform manual labor [11]. Positive aging can let workers work as long as possible while ensuring improved health and safety [12]. Construction workers with positive MHE are better able to perceive the risks of the construction site and improve their SV. Construction workers with negative MHE may feel stressed due to age sensitivity, which may reduce their sense of security and job satisfaction [55]. When a lot of psychological stress accumulates, common mental disorders such as depression, anxiety, and adjustment disorder will follow [56], the risk in the environment will not be well perceived, and SA will be affected. Therefore, it can be inferred that MHE has positive significance for SRP and SA. And the following hypotheses are made:

Hypothesis c3 (Hc3):

MHE has positive significance for SRP.

Hypothesis c4 (Hc4):

MHE has positive significance for SA.

2.4.3. CFE and SV

Cognition, as a bodily function [57], can be defined as the ability to acquire and process information, apply knowledge, and use experience to solve problems [58]. In the process of aging, cognitive function declines with age, which is a widespread syndrome [59]. As the population ages rapidly, the burden of diseases related to cognitive function increases [60], which affects people’s quality of life. Therefore, people’s expectations about cognitive function can influence their future health outcomes [61] and behaviors [62]. Construction workers have positive CFE, so their memory and judgment skills are better. They are better able to identify security risks [63]. Construction workers who have negative CFE also have negative SA and poor judgment skills [63]. Therefore, it can be inferred that CFE has positive significance for SRP and SA. And the following hypotheses are made:

Hypothesis c5 (Hc5):

CFE has positive significance for SRP.

Hypothesis c6 (Hc6):

CFE has positive significance for SA.

Combining the above hypotheses, this study establishes a mediation model to investigate the mediating mechanism of AE on CWSB. This model aims to provide a comprehensive understanding of the factors and psychological mechanisms influencing CWSB, thereby offering a scientific foundation for designing effective interventions. The hypothesized conceptual framework is shown in Figure 1.

3. Method

3.1. Data Collection

3.1.1. Survey Instrument

This study used a structured questionnaire developed in previous research to collect CWSB data. CWSB was evaluated based on compliance with rules and regulations and participation in safety training.

The questionnaire was divided into four parts. In the first part, the basic information of the respondents, such as education level, working years, etc., was collected to ensure the effectiveness of the collected data. The next two sections were about AE from three dimensions, namely PHE, MHE, and CFE, and SV from two dimensions, namely SRP and SA. The last section was about four items of CWSB, as shown in the following Table 1.

In the assessment of construction safety, the variables of CWSB, AE, and SV were quantified using a 5-point Likert scale [64]. This scale is prevalent in construction safety research. Each item specifically evaluated whether construction personnel execute tasks correctly and safely. The scale was structured such that a score of 1 indicates ‘strongly disagree’, while a score of 5 signifies ‘strongly agree’.

Table 1. Definitions and sorted items for constructs.

AE	PHE	When people get older, they need to lower their expectations of how healthy they can be.	[18,65]
		The human body is like a car: When it gets old, it gets worn out.
		Having more aches and pains is an accepted part of aging.
		Every year that people age, their energy levels go down a little more.
	MHE	I expect that I get older I will spend less time with friends and family.	[18,65]
		Being lonely is just something that happens when people get old.
		As people get older they worry more.
		It’s normal 10 be depressed when you are old.
	CFE	I expect that as I get older I will become more forgetful.	[18,65]
		It’s an accepted part of aging to have trouble remembering names.
		Forgetfulness is a natural occurrence just from growing old.
		It is impossible to escape the mental slowness that happens with aging.
SV	SRP	My skills, experiences and knowledge enable me to control safety risks and get my job done.	[66,67,68]
		Management turns a blind eye to my risk-taking behaviour.
		My colleagues support my risk-taking behaviour as I get my job done.
		I can identify hazards and hidden dangers in my workplace.
	SA	I know the importance of safety in the construction industry.	[69,70]
		I am always concerned about safety issues in production operations.
		I work especially carefully when I’m not sure it’s safe.
CWSB	I don’t pay attention to safety practices and regulations.		[46,71]
	I have a problem with being involved in creating a safe and civilized site.
	I do not strictly follow safe operating procedures and regulations.
	I will be too familiar with the work to ignore safety practices.

Table 1. Definitions and sorted items for constructs.

AE	PHE	When people get older, they need to lower their expectations of how healthy they can be.	[18,65]
		The human body is like a car: When it gets old, it gets worn out.
		Having more aches and pains is an accepted part of aging.
		Every year that people age, their energy levels go down a little more.
	MHE	I expect that I get older I will spend less time with friends and family.	[18,65]
		Being lonely is just something that happens when people get old.
		As people get older they worry more.
		It’s normal 10 be depressed when you are old.
	CFE	I expect that as I get older I will become more forgetful.	[18,65]
		It’s an accepted part of aging to have trouble remembering names.
		Forgetfulness is a natural occurrence just from growing old.
		It is impossible to escape the mental slowness that happens with aging.
SV	SRP	My skills, experiences and knowledge enable me to control safety risks and get my job done.	[66,67,68]
		Management turns a blind eye to my risk-taking behaviour.
		My colleagues support my risk-taking behaviour as I get my job done.
		I can identify hazards and hidden dangers in my workplace.
	SA	I know the importance of safety in the construction industry.	[69,70]
		I am always concerned about safety issues in production operations.
		I work especially carefully when I’m not sure it’s safe.
CWSB	I don’t pay attention to safety practices and regulations.		[46,71]
	I have a problem with being involved in creating a safe and civilized site.
	I do not strictly follow safe operating procedures and regulations.
	I will be too familiar with the work to ignore safety practices.

3.1.2. Questionnaire Survey

Questionnaires were distributed directly on-site, and each item was thoroughly explained to construction workers prior to completion. The selection of cities for this survey was based on their robust economic performance, as indicated by GDP data. Specifically, Wuhan in Hubei Province, Changsha in Hunan Province, and Hefei in Anhui Province were chosen for their high levels of economic development. Notably, in 2022, Anhui Province’s construction sector experienced the second-highest year-on-year GDP growth nationally. Hefei, which accounts for half of Anhui’s construction GDP, demonstrates significant developmental momentum. These cities were selected due to their advanced construction industry development. Anhui Province, in particular, stands out for the professional development opportunities it offers to construction workers. This strategic sample selection ensured that the survey accurately captured the intended perspectives. The study ultimately collected 457 responses from construction workers across various job categories.

3.2. Data Analysis

3.2.1. Data Screening

Out of the 457 questionnaires initially distributed on-site, adjustments were made for data quality: 45 were excluded due to incomplete responses, 12 due to the selection of multiple answers per item, and 9 for responses that deviated significantly from the intended question meanings. Consequently, 391 questionnaires were deemed valid, yielding an effective response rate of 85.6%. This rate aligns with the requirements for adequate sample size and demographic representation. Regarding structural equation modeling (SEM), it is recommended that the minimum sample size should exceed 10 observations per indicator [72], without a specified upper limit [73]. Given that the proposed model includes 26 indicators, the sample of 391 valid questionnaires is appropriate and expected to yield high-quality estimates [74].

3.2.2. Data Analysis

Statistical analyses were conducted using IBM SPSS Statistics 24 and Amos 28. Initially, an internal consistency reliability test was performed to assess the cohesion among the questionnaire variables. A high reliability coefficient is commonly interpreted as indicative of the items’ ability to measure an underlying construct effectively. Subsequent analyses involved confirmatory factor analysis (CFA) and structural equation modeling (SEM) to test the proposed hypotheses. CFA evaluated the model’s reliability and validity, encompassing both convergent and discriminant validity. Convergent validity assesses how well individual scale items reflected a common construct, whereas discriminant validity examines the distinctiveness of the constructs—specifically, the three dimensions of AE, SV, and CWSB. The adequacy of the factorial structure of these constructs was also consistently presented. SEM was employed to validate the research hypotheses by exploring the interrelationships among the different constructs.

4. Results

4.1. Demographics of the Respondents

Participants ranged in age from 18 to 60 years, with 87.7% reporting being married. Concerning educational attainment, 28.6% of respondents held a high school diploma or a lower qualification. The work experience among participants varied from 1 to 26 years. Demographic details are presented in

Table 2. Statistical results of demographic and working characteristics (N = 391).

Variable		Number	Percentage
Gender	Male	338	86.4%
	Female	53	13.6%
Age (years)	18–25	14	3.6%
	26–33	57	14.6%
	34–44	117	29.9%
	45–55	150	38.4%
	≥56	52	13.3%
Educational attainment	Primary and below	67	17.1%
	Junior high school	171	43.7%
	High school/ technical secondary school	112	28.6%
	College degree, bachelor’s degree or above	36	9.2%
Marital status	Single	37	9.5%
	Married	343	87.7%
	Divorced or widowed	11	2.8%
Work experience (years of work)	≤5 years	45	11.5%
	6–10 years	90	23.0%
	11–15 years	102	26.1%
	16–20 years	63	16.1%
	≥21 years	89	22.8%

.

4.2. Scale Reliability and Validity Test

Analyzing the reliability of constructs used in structural equation modeling is a crucial step to ensure the consistency and reliability of measurement instruments. Generally, the higher the reliability, the more accurate and dependable the data. The Cronbach’s Alpha coefficient is commonly used to evaluate the reliability of data collected using a Likert scale. This coefficient ranges from 0 to 1, with a value above 0.7 indicating that the sample data obtained from the questionnaire has good reliability [75]. According to the data presented in

Table 3. Reliability analysis.

Variables	Item Number	Cronbach’s Alpha	Overall Cronbach’s Alpha
PHE	4	0.891	0.895
MHE	4	0.875
CFE	4	0.826
SRP	4	0.816
SA	3	0.766
CWSB	4	0.832

, each dimension within this research demonstrated high reliability.

Before conducting the validity analysis, it is essential to test two indicators: the Kaiser–Meyer–Olkin (KMO) value and Bartlett’s test of sphericity. These tests must be passed before proceeding with the validity analysis. The conditions are satisfied when the KMO value is greater than 0.6 and Bartlett’s test results in a p-value less than 0.001. According to the data in

Table 4. KMO values and Bartlett’s test of sphericity results.

Kaiser–Meyer–Olkin Measure of Sampling Adequacy		0.875
Bartlett’s test of sphericity	Approx. chi-square	4380.445
	df	253
	Sig.	0

, the KMO value was 0.875, indicating that the sampling was adequate for factor analysis. Additionally, the Bartlett’s test yielded a significant chi-square statistic with a p-value of 0.000. These results confirmed that the index was appropriately suited for factor analysis.

The structural equation model consists of multiple aspects, each containing a latent variable and multiple observed variables. Before conducting the path analysis of the structural equation, it is essential to examine each aspect for its composite reliability (CR) and convergent validity. A CR greater than 0.7 [76] indicates that the observed variables represent the latent variable well, ensuring the stability of the questionnaire. Convergent validity, also known as average variance extracted (AVE), assesses whether the observed variables effectively reflect the latent variables they are intended to measure. When the AVE is higher than 0.5 [2], it is considered satisfactory, indicating that the construct has sufficient convergent validity.

Table 5. Convergent validity analysis.

Dimension	Variable	Mean	SD	Convergent Validity Standardized Regression Weights	CR	AVE
PHE	PHE1	4.23	0.948	0.749	0.892	0.675
	PHE2	4.09	0.941	0.889
	PHE3	3.96	0.944	0.783
	PHE4	4.10	0.909	0.857
MHE	MHE1	3.93	0.865	0.767	0.876	0.639
	MHE2	3.88	0.930	0.829
	MHE3	3.89	0.912	0.825
	MHE4	3.98	0.827	0.775
CFE	CFE1	4.37	0.735	0.686	0.829	0.549
	CFE2	4.42	0.780	0.789
	CFE3	4.23	0.871	0.716
	CFE4	4.36	0.782	0.766
SRP	SRP1	4.79	0.561	0.740	0.826	0.543
	SRP2	4.76	0.594	0.773
	SRP3	4.54	0.746	0.688
	SRP4	4.65	0.742	0.743
SA	SA1	4.50	0.744	0.798	0.776	0.538
	SA2	4.41	0.845	0.709
	SA3	4.31	0.979	0.688
CWSB	CWSB1	4.15	0.984	0.756	0.832	0.554
	CWSB2	4.30	0.897	0.765
	CWSB3	4.25	0.960	0.736
	CWSB4	4.08	1.006	0.719

CR = composite reliability and AVE = average variation extraction.

showed that all standardized loading coefficients were significant and all CR indices exceeded 0.7, indicating that the observed variables explained the latent variables well. The AVE values of all six factors exceeded 0.5, with the AVEs of PHE and MHE even exceeding 0.6, thus confirming the strong convergent validity of the research.

Discriminant validity was employed to assess the extent of differentiation between constructs. It was evaluated using the Fornell–Larcker criterion, which compares the square root of the AVE values to the correlations of the underlying constructs. Discriminant validity was confirmed when the square root of the AVE for each factor exceeded the correlation coefficients of the other factors in the same column. According to the data presented in

Table 6. Discriminant validity analysis.

	AVE	PHE	MHE	CFE	SRP	SA	CWSB
PHE	0.675	0.822
MHE	0.639	0.339	0.799
CFE	0.549	0.311	0.383	0.741
SRP	0.543	0.287	0.206	0.310	0.737
SA	0.538	0.332	0.337	0.436	0.183	0.733
CWSB	0.554	0.510	0.462	0.542	0.403	0.439	0.744

AVE = average variation extraction.

, the discriminant validity across constructs satisfied the established criteria. This indicated that the inter-construct correlations were sufficiently low, validating the use of SEM for the analysis. The findings demonstrated robust discriminant validity, affirming the suitability of the model for detailed SEM analysis.

4.3. Structural Equation Model Fit Test

CMIN/DF assesses model fit by comparing the observed covariance matrix with the estimated covariance matrix. Ideal values between 1 and 3 indicate a good fit, suggesting that the model is reasonable. RMSEA measures how well the model fits the overall covariance matrix when the parameter estimates are unknown but optimally chosen. An RMSEA value of less than 0.05 indicates that the model fits the degrees of freedom closely. The GFI evaluates the ratio of covariance explained by the model to the covariance of the actual observed data. Values greater than 0.9 are generally considered to indicate a good fit. NFI measures the improvement in fit between the hypothetical model and the null model. A value greater than 0.9 indicates a good fit. CFI compares the degree of fit between the hypothetical model and the independent model. A value greater than 0.9 indicates a good fit. Similar to CFI, IFI values greater than 0.9 indicate a good fit. PGFI adjusts GFI according to the number of parameters in the model, and values greater than 0.5 indicate a good fit. PNFI adjusts NFI according to the complexity of the model, and values greater than 0.5 indicate a good fit. The results shown in

Table 7. Model fit evaluation form.

Index	Absolute Fit			Value-Added Fit Index			Parsimonious Fit Index
	CMIN/DF	RMSEA	GFI	NFI	CFI	IFI	PGFI	PNFI
Ideal value	(1,3)	<0.05	>0.9	>0.9	>0.9	>0.9	>0.5	>0.5
Actual value	1.888	0.048	0.918	0.909	0.955	0.955	0.718	0.776
Judge	Fit	Fit	Fit	Fit	Fit	Fit	Fit	Fit

CMIN/DF = chi-square/degrees of freedom; RMSEA = root-mean-square error of approximation; GFI = goodness-of-fit index; NFI = normative fit index; CFI = comparative fit index; IFI = incremental fit index; PGFI = parsimony goodness-of-fit index; PNFI = parsimony-adjusted normed-fit index.

indicated that the model fitted the data well, providing support for the path analysis results and the proposed theoretical relationships.

4.4. Structural Equation Model Path Validation

After the confirmation of fitness, the structural equation model is analyzed as a whole, as shown in Figure 2. The standardized path coefficients between the latent variables can be clearly observed, with specific analysis results presented in

Table 8. Model estimation result.

Hypothesis	Path	Estimate	S.E.	C.R.	p	Result
Ha1	PHE → CWSB	0.259	0.054	4.785	***	Supported
Ha2	MHE → CWSB	0.207	0.062	3.322	***	Supported
Ha3	CFE → CWSB	0.330	0.076	4.362	***	Supported
Hb1	SRP → CWSB	0.312	0.094	3.321	***	Supported
Hb2	SA → CWSB	0.166	0.073	2.269	0.023 *	Supported
Hc1	PHE → SRP	0.112	0.035	3.180	0.001 **	Supported
Hc2	PHE → SA	0.145	0.049	2.944	0.003 **	Supported
Hc3	MHE → SRP	0.033	0.042	0.791	0.429	Unsupported
Hc4	MHE → SA	0.142	0.059	2.381	0.017 *	Supported
Hc5	CFE → SRP	0.158	0.047	3.396	***	Supported
Hc6	CFE → SA	0.319	0.067	4.784	***	Supported

S.E. = standard error; C.R. = composite reliability. * Correlation is significant at 0.05 level. ** Correlation is significant at 0.01 level. *** Correlation is significant at 0.001 level.

. Except for the path MHE → SRP, the other 10 hypothesized paths between the latent variables (PHE → CWSB, MHE → CWSB, CFE → CWSB, SRP → CWSB, SA → CWSB, PHE → SRP, PHE → SA, MHE → SA, CFE → SRP, and CFE → SA) are significant at the $p<0.05$ level and all are positive. Furthermore, the paths PHE → CWSB, MHE → CWSB, CFE → CWSB, SRP → CWSB, CFE → SRP, and CFE → SA are significant at the $p<0.001$ level. This indicated that all ten hypothesized paths in the model were valid and had a significant positive effect on CWSB. These results suggested that AE and SV can effectively influence CWSB, demonstrating that the model was highly adaptable and relevant within this research context.

4.5. Analysis of Results

Among the three dimensions of AE, namely PHE, MHE, and CFE, PHE and CFE exert an equal degree of influence on CWSB, with standardized path coefficients of 0.27. The main perceptions of CWSB by PHE of construction workers come from four main areas: lowering expectations of health during the aging process, decreasing physical functioning, suffering from illness, and decreasing energy. The factor loading coefficients of these four observed variables were 0.75, 0.89, 0.78, and 0.86, indicating that CWSB was more affected by physical functioning during aging. The higher the expectation of physical functioning, the stronger the CWSB. The factor loading coefficients of the four observed variables of CFE were 0.69, 0.72, 0.79, and 0.77, respectively, indicating that the higher the CFE of the construction workers, the more inclined they were to perform CWSB. MHE had a lesser effect on CWSB than PHE and CFE, with a standardized path coefficient of 0.18. This indicated that MHE had a greater effect on CWSB in four main influences: lack of family and friends, fear of loneliness, worrying about more things, and depression.

Of the two dimensions of SV, SRP and SA, SRP has a greater degree of influence on CWSB, with a standardized path coefficient of 0.18. The influence of construction workers’ SRP in their CWSB is mainly reflected in the following four aspects: the construction workers’ own experience, skills, and knowledge; the management’s disregard for the construction workers’ adoption of risk-taking behaviors; the friends of the construction workers’ support for the adoption of risk-taking behaviors; and the construction workers’ own ability to recognize hazardous hazards in the workplace. The factor loading coefficients of these four observed variables are 0.74, 0.77, 0.69, and 0.77, respectively, indicating that management’s attitude towards whether construction workers adopt CWSB is more important. Construction workers will adopt CWSB if the management strictly stops their risky behavior and takes appropriate punitive measures against them. The standardized path coefficient of SA on SB is 0.14, indicating that the chance of construction workers adopting CWSB increases when they have good SA.

PHE positively affects SRP and SA with standardized path coefficients of 0.20 and 0.18, respectively. This suggests that the SRP and SA of construction workers are more affected by the observed variable of physical functioning during aging. The higher the construction worker’s expectation of physical functioning, the stronger their SRP and SA. CFE positively affects SRP and SA with standardized path coefficients of 0.23 and 0.32, respectively, which suggests that the higher the construction worker’s expectation of CFE, the higher their perception of risk and the stronger the effect on their attitudinal change towards working safely. MHE positively affects SA with a standardized path coefficient of 0.15. The higher the construction worker’s MHE, the higher the construction worker’s perception of risk and the stronger the effect on their attitudinal change towards working safely. Workers will adopt safe work practices when their MHE is higher. The standardized path coefficient of MHE on SRP is 0.05, which means that the MHE of construction workers has no effect on SRP.

4.6. Mediation Effect Analysis

Indirect effects were indicative of the presence of mediating variables within the linkage framework.

Table 9. Mediation effect test of the model.

		Product of Coefficient Multiplication		Bootstrapping
				Bias-Corrected 95%CI		Percentile 95%CI
Path	Point Estimate	SE	Z	Lower	Upper	Lower	Upper
PHE→SRP→CWSB	0.035	0.025	1.400	0.007	0.107	0.007	0.103
PHE→SA→CWSB	0.024	0.015	1.600	0.003	0.065	0.001	0.061
MHE→SRP→CWSB	0.010	0.016	0.625	−0.013	0.049	−0.015	0.048
MHE→SA→CWSB	0.024	0.017	1.412	0.002	0.077	0.000	0.064
CFE→SRP→CWSB	0.049	0.028	1.750	0.014	0.135	0.010	0.122
CFE→SA→CWSB	0.053	0.029	1.828	0.010	0.122	0.003	0.111
Total	0.195	0.067	2.910	0.077	0.341	0.081	0.347

presents the effect coefficients (standardized values), detailing the levels of impact among various variables. Notably, since AMOS software reports only the total indirect effects from one variable to another [77], additional coding was implemented to extract specific distal mediation effects. The confidence intervals not encompassing zero suggest that CWSB is significantly influenced by AE through SV, thereby confirming the presence of a significant mediating pathway.

5. Discussion

The aging workforce has become a serious global issue [17], particularly in the construction industry, where aging is associated with declining physical health and an increase in chronic health problems [60]. Our study differs from previous research on factors influencing CWSB by exploring the effects of AE on CWSB. Our empirical results confirm a positive correlation between AE and CWSB. Compared to workers with negative AE, those with positive AE exhibit more CWSB, consistent with prior findings on the influence of AE on CWSB [78]. Importantly, our study adopts a positive psychology perspective, subdividing AE into three dimensions: PHE, MHE, and CFE. Each dimension’s relationship with CWSB is explored separately. This approach validates the role of AE in promoting CWSB and extends it as a key factor representing individual cognition in the field of safety. By establishing AE as an individual-level predictor of CWSB, our research reveals a positive psychology perspective previously overlooked in CWSB literature, providing valuable insights and contributing positively to the current research on human factors and safety.

Furthermore, our study finds that SV mediates the positive relationship between AE and CWSB, offering deeper insights beyond a direct relationship. While previous studies have independently validated the positive correlations between AE and CWSB [78], and between SV and CWSB [21], the relationship among these three factors was unclear. By applying SVT, we integrate AE, SV, and CWSB into a unified framework, confirming the mediating role of SV in the AE-CWSB relationship. Interestingly, we found that the positive impact of MHE on the SRP component of SV was not confirmed. This may be because the subjective intentions regarding MHE among construction workers have a limited effect on their SRP. Existing research suggests that factors such as attention [40] have a more significant influence on SRP. Studies using eye-tracking technology have shown that cognitive processes dominating attention can enhance SRP through the analysis of eye movements and visual inspections. When construction workers are attentive, they are more likely to accurately identify potential safety hazards and take preventive measures [42]. Conversely, a lack of focus may lead to the failure to recognize or underestimate safety risks, resulting in poor safety decisions and an increased likelihood of accidents [44].

From the SVT perspective, while cognitive factors (AE) are recognized as drivers for adhering to safety protocols, this study’s limitation lies in its cross-sectional design. The existing literature lacks longitudinal studies, making it challenging to examine the long-term sustainability of the causal relationship between AE and CWSB. Additionally, individuals with lower formal education levels might struggle to select Likert-type responses accurately, potentially affecting the validity of the responses. The skewness in variables such as education could be a reason for decreased correlations. Research on age expectation has predominantly focused on younger populations, while the aging population in the construction industry has often been marginalized in academic investigations. This oversight neglects the physiological and cognitive dynamics that change with age, which are crucial for adapting to existing safety measures. Moreover, the concept of SV is often approached from a top-down perspective, prioritizing organizational policies while overlooking deep-rooted cultural and social factors influencing individual construction workers. The intersection of individual SV and collective organizational ethos requires more detailed exploration. In conclusion, there is an urgent need to expand current academic research to comprehensively and sustainably enhance CWSB. Longitudinal studies should be conducted to measure the sustained impact of SVT. Additionally, future research must include underrepresented older workers and consider the nuances of grassroots SV to develop more comprehensive and adaptable safety paradigms.

6. Conclusions

With authorization from the construction industry, we collected 391 valid questionnaires from construction workers across various Chinese cities. This research validated most hypotheses through descriptive statistics, correlation analysis, structural equation modeling, multiple fit indices, and analysis of common method variance. However, the hypothesis positing a significant positive impact of MHE on SRP was not supported. The findings indicate that AE directly influences CWSB, whereas SV exerts an indirect influence. Furthermore, this study makes significant contributions to the existing body of knowledge in several key areas: Firstly, it examines the sensitivity of construction workers to age-related changes and how their AE might CWSB amid rapid demographic shifts. Previous research has largely overlooked the direct insights of construction workers on how age-related changes impact their safety behaviors. By collecting data directly from construction workers, this study provides a clear visualization of the AE among construction workers of different ages, enhancing the theoretical and practical foundations of management, medicine, psychology, and safety science. Secondly, the study incorporates the construct of SV, categorized under personal values. Prior studies scarcely explored the empirical relationship between SV, AE, and CWSB. This study addresses this omission, offering a comprehensive understanding of how AE and SV relate to CWSB. Such insights are crucial for improving management practices and theoretical approaches within construction enterprises, ultimately fostering a safer industry environment.