Cognitive Bias and Unsafe Behaviors in High-Altitude Construction Workers Across Age Groups

Abstract

The present study aims to investigate how the safety cognition bias of high-altitude workers in different age groups in the construction industry affects their unsafe behaviors. A theoretical framework involving safety cognition bias, risk-taking propensity, work experience, and unsafe behaviors was constructed. The empirical analysis was conducted using structural equation modeling. The results indicate that safety cognition bias has a significant positive effect on the unsafe behaviors of high-altitude workers. Risk-taking propensity plays a mediating role in this relationship, with partial mediation for the new generation of high-altitude workers and full mediation for the older generation. Moreover, work experience plays a crucial role in moderating the relationship between safety cognition bias and unsafe behaviors, specifically showing a significant moderating effect on the new generation of high-altitude workers, while its impact on the older generation is not significant. These findings provide theoretical support and practical guidance for reducing unsafe behaviors in high-altitude construction work, enhancing the safety management level of construction workers.

1. Introduction

According to data statistics from the Ministry of Housing and Urban–Rural Development of China, from 2019 to 2023, a total of 3362 production safety accidents occurred in China’s housing and municipal engineering industries, resulting in 3813 fatalities. These accidents have caused severe casualties and property losses in construction production. Among these accidents, classified by type, falls from height dominated the construction industry, with a total of 1785 cases, accounting for as much as 53.09%, making high-altitude workers the group most prone to accidents [1,2]. High-altitude workers are particularly vulnerable to accidents due to the nature of their work, which involves significant risks such as falls from height. The data indicate that falls from height are the leading cause of fatalities in the construction industry, accounting for over half of all accidents. This underscores the importance of focusing on high-altitude workers as a key group in safety research. Additionally, the increasing number of young, highly educated newcomers in the high-altitude operation field has introduced new trends of generational alternation and technological innovation. These newcomers, compared to the older generation, show significant differences in safety cognition, often lacking sufficient work experience and safety knowledge, which makes them more prone to cognitive biases and unsafe behaviors. Therefore, understanding and addressing the cognitive biases of high-altitude workers is crucial for enhancing safety management in the construction industry.

The causes of falls from height cover multiple domains, including improper personnel behavior, improper handling of materials and equipment, environmental safety hazards, and inadequate implementation of management measures [3,4]. Research indicates that unsafe behaviors of personnel are the primary cause of accidents, and safety cognition bias is the deep-rooted source of such improper behaviors. In recent years, the high-altitude operation field in China’s construction industry has shown new trends of generational alternation and technological innovation [5]. The structure of high-altitude operation personnel has changed, with an increasing number of young, highly-educated, and strong-learning-ability newcomers. However, compared with the older generation of high-altitude operation personnel, the newcomers show significant differences in safety cognition. Due to their lack of sufficient work experience and the accumulation of safety knowledge, they are more likely to become victims of falls from height accidents [6].

Comparing the safety cognition and behaviors of young and older high-altitude workers is essential for several reasons. First, generational differences in values, attitudes, and behaviors can significantly affect workers’ understanding and compliance with safety regulations. Younger workers, often with less experience and exposure to hazardous situations, may have a higher propensity for cognitive biases and risk-taking behaviors. In contrast, older workers, with their accumulated experience, may have a more accurate perception of risks and a stronger adherence to safety protocols. Second, understanding these generational differences can help tailor safety training and interventions to address the specific needs of each group. For example, targeted training programs for younger workers can focus on enhancing situational awareness and reinforcing safety protocols, while leveraging the experience of older workers through mentorship programs can help transfer critical safety knowledge. Finally, identifying the mediating and moderating factors that influence the relationship between cognitive biases and unsafe behaviors across different age groups can provide valuable insights for improving overall safety management in the construction industry.

While the safety cognition bias model has been widely used to explain unsafe behaviors in the construction industry, several critiques and counterarguments have been raised. Some scholars argue that the model may be too subjective and culturally biased, potentially leading to misinterpretations when applied across different cultural contexts. Additionally, the model has been criticized for oversimplifying individual differences and neglecting the dynamic nature of construction environments. These critiques highlight the need for a more nuanced understanding of safety behaviors that incorporates both individual and contextual factors.

To address these concerns, this study incorporates additional variables such as work experience and risk-taking propensity into the theoretical framework. By considering these factors, we aim to provide a more comprehensive understanding of the relationship between safety cognition bias and unsafe behaviors.

In light of this, this paper will start from the perspective of generational differences and delve into the intrinsic connection between safety cognition bias and the unsafe behaviors of high-altitude operation personnel.

In construction sites, high-altitude workers face complex and variable activities and environmental factors. These external conditions, combined with differences in individual social backgrounds and experiences, shape their unique perspectives on work cognition. These perspectives can become potential factors inducing unsafe behaviors [7,8]. Some scholars have pointed out that when facing dangerous situations, if workers fail to take appropriate measures and instead choose unsafe behaviors, this is often attributed to a numb psychological state. This psychological state can prompt individuals to take risks impulsively and ignore potential dangers, thereby increasing the risk of accidents [9]. Based on this view, we can preliminarily hypothesize that the impact of safety cognition bias on unsafe behaviors may be realized by stimulating individuals’ risk-taking tendencies. In addition, relevant scholars have conducted experimental studies to explore the changes in the hazard identification performance of construction workers with different experience levels after receiving expert hazard identification eye-movement model examples (EMMEs) intervention. The study used an eye-tracker to record the eye-movement data of high-and low-experience groups of workers in identifying hazards in different scenarios, and assessed the performance changes across three dimensions: identification accuracy, completion time, and the standardization of identification sequence [10,11]. The results showed that the EMMEs intervention group’s hazard identification performance was significantly higher than that of the non-intervention group; in the pre-intervention test, the high-experience group’s identification performance was far better than that of the low-experience group; and the EMMEs intervention effect had a professional knowledge reversal effect, with the low-experience group’s identification performance improving significantly more after intervention than the high-experience group [12]. This indicates that work experience plays an important role in regulating workers’ cognition of safety hazards, thereby affecting the relationship between safety cognition bias and unsafe behaviors. Although existing studies have recognized the impact of safety cognition bias on unsafe behaviors, the exploration of its internal structural relationship is still insufficient. At the same time, there is a lack of systematic comparative research on the causes of behavior intentions of high-altitude workers of different age groups in the construction industry.

Given the significant impact of safety cognition bias on unsafe behaviors and the potential differences between generations, this study aims to address the following research questions:

• How does safety cognition bias influence the unsafe behaviors of high-altitude workers across different age groups?
• What role does risk-taking propensity play in mediating the relationship between safety cognition bias and unsafe behaviors?
• How does work experience moderate this relationship, particularly among younger and older generations of high-altitude workers?

To address these questions, we construct a theoretical model incorporating safety cognition bias, risk-taking propensity, work experience, and unsafe behaviors. This study aims to provide a comprehensive understanding of the underlying mechanisms and offer practical insights for improving safety management in the construction industry.

The findings from this study have significant implications for practical construction safety training. By identifying the role of safety cognition bias and risk-taking propensity in unsafe behaviors, we can develop targeted training programs that address the specific needs of different age groups. For the younger generation of high-altitude workers, training should focus on enhancing situational awareness, reinforcing safety protocols, and building a strong foundation of safety knowledge. For older workers, leveraging their extensive experience through mentorship programs can help transfer critical safety knowledge to younger workers while maintaining adherence to safety regulations. Additionally, incorporating advanced technologies such as virtual reality (VR) and artificial intelligence (AI) into training programs can provide immersive experiences that enhance workers’ hazard recognition abilities and situational awareness. These tailored interventions can significantly reduce the incidence of accidents and improve overall safety performance in the construction industry.

2. Theoretical Foundations and Research Hypotheses

2.1. Generational Differences

Generational differences are characterized by distinct values and behavioral patterns among individuals from different birth cohorts, influenced by diverse sociocultural environments [13]. These differences affect workers’ understanding and compliance with safety regulations, leading to varied behavioral manifestations [14,15]. Thus, examining these differences is essential for preventing unsafe behaviors.

The differences in safety cognition and behavior among generations can be attributed to several factors. First, sociocultural environments have evolved over time, influencing the values and attitudes of different generations. For example, younger workers are more likely to be influenced by rapid technological advancements and a greater emphasis on individualism, which may lead to higher risk-taking propensity compared to older workers who value traditional safety norms more [16]. Second, the level of education and access to information also vary between generations. The new generation of workers typically has higher educational attainment and exposure to modern safety training, which may shape their understanding of safety risks differently than the older generation [17,18]. Lastly, the work experience accumulated over the years by older workers provides them with a deeper understanding of the consequences of unsafe behaviors, making them more cautious and less prone to cognitive biases [19].

Based on the overall sample analysis, this study adopts the age classification standards of the World Health Organization (WHO), defining individuals under 44 years old as young, those aged 45 to 59 as middle-aged, and those aged 60 and above as elderly. Additionally, with 1 January 1980 as the dividing line (with 2024 as the reference year), construction high-altitude workers are further categorized into the new generation and the older generation [20,21]. This classification aims to precisely grasp and comparatively analyze the differences in safety cognition between these two generational groups, as well as how these cognitive biases influence their behavioral intentions. Such a division facilitates a deeper understanding of the specificity of safety cognition across generations and its impact on operational safety.

2.2. Safety Perception Bias and Unsafe Behavior

Safety cognition bias is defined as the systematic deviation in individuals’ perceptions, judgments, and assessments of safety-related risks and hazards. This bias occurs when workers’ subjective evaluations of safety risks do not align with objective safety standards or established protocols. Such deviations can lead to underestimation of potential dangers, overconfidence in personal abilities, or misjudgment of the effectiveness of safety measures [22]. For example, a worker might underestimate the risk of falling from a height due to overconfidence in their experience or skills, leading to unsafe behaviors such as skipping safety checks or not wearing personal protective equipment.

Extensive research has established a positive causal relationship between safety cognition bias and unsafe behaviors, with cognitive bias being a key factor in workers’ unsafe actions [23,24]. For example, recent studies have shown that cognitive biases can exacerbate adverse mental states under multidimensional psychological stressors, further increasing the likelihood of unsafe behaviors. Additionally, risk perceptions have been identified as a mediating factor in this relationship, highlighting the importance of addressing cognitive biases to enhance safety management.

Recent studies have continued to explore the role of safety cognition bias in the construction industry, highlighting its dynamic nature and evolving impact. For instance, one study used EEG-based detection to analyze the adverse mental states of construction workers at height, revealing that cognitive biases can be exacerbated by multidimensional psychological stressors [25]. Another study investigated the mediating effects of risk perceptions on the relationship between cognitive biases and unsafe behaviors among tunnel construction workers, emphasizing the importance of addressing cognitive biases to enhance safety management [26]. These findings underscore the need for continuous updates in safety training programs to address the evolving nature of cognitive biases in high-risk environments.

Moreover, the integration of advanced technologies such as virtual reality (VR) and artificial intelligence (AI) has been proposed to simulate realistic construction scenarios and improve workers’ hazard recognition abilities. For example, a recent empirical study suggested that VR-based training could effectively reduce cognitive biases by providing immersive experiences that enhance workers’ situational awareness and decision-making skills [27]. These technological advancements highlight the potential for innovative solutions to mitigate the impact of cognitive biases on unsafe behaviors in the construction industry.

To further elucidate this relationship, we draw on the “Dual-Process Theory (DPT)”, which provides a robust framework for understanding how cognitive biases influence safety behaviors. According to DPT, human cognition operates through two distinct systems: the automatic system, which is fast, intuitive, and prone to biases; and the reflective system, which is slower and more deliberate. In high-altitude work environments, the automatic system often dominates due to time constraints and complex tasks, making workers more susceptible to cognitive biases. These biases can lead to underestimation of risks and overestimation of personal abilities, thereby increasing the likelihood of unsafe behaviors. Younger workers, with less experience to draw upon, are particularly vulnerable to these biases, while older workers, with their accumulated experience, are better equipped to engage the reflective system and mitigate these biases. This generational difference underscores the need for targeted safety interventions.

Given the differences in the age composition of high-altitude workers in the construction industry, their safety thinking, values, and behavioral habits also vary, which further affects their accuracy in perceiving safety risks [28]. Specifically, the new generation of high-altitude workers, due to the lack of necessary experience and professional skills, often find it difficult to fully and accurately identify potential risks and hazards, thus being more prone to safety cognition bias. In contrast, the older generation of high-altitude workers, with their years of work experience, have a deeper understanding and experience of the consequences of unsafe behaviors, and their safety cognition is relatively more accurate [29,30].

Given this theoretical framework, we propose the following hypothesis:

H1. 

Safety cognition bias has a positive impact on unsafe behaviors, and this impact varies significantly among high-altitude workers of different age groups.

2.3. The Moderating Role of Work Experience

In examining the relationship between safety cognition bias and unsafe behaviors, it is essential to consider the role of work experience as a potential moderator. Work experience is not merely a measure of time spent in a profession; it represents the accumulation of knowledge, skills, and situational awareness that workers develop over their careers. This accumulation of experience can profoundly influence how workers perceive and respond to safety risks, thereby affecting the manifestation of unsafe behaviors [31].

The basic principle underlying the moderating role of work experience lies in the dual-process theory (DPT) of human cognition, which posits that cognitive processes can be divided into two systems: the automatic system, which is fast and intuitive but prone to biases; and the reflective system, which is slower and more deliberate. In high-altitude construction work, where tasks are often complex and time-sensitive, workers frequently rely on the automatic system for decision-making. However, experienced workers, through repeated exposure to hazardous situations and the consequences of unsafe behaviors, develop a more refined reflective system. This enhanced reflective system allows them to better recognize and counteract cognitive biases, leading to safer behaviors

Experts, through in-depth analysis of unsafe behavior data, have confirmed that work experience is one of the key factors affecting the frequency of unsafe behaviors [32]. By constructing a theoretical cognitive model, it has been observed that compared to employees with little experience, experienced employees have significant differences in identifying dangers and can more accurately perceive potential risks. At the same time, new employees tend to trust and imitate experienced senior employees to jointly create a safe working environment [33,34]. This suggests that experience-based training and mentorship programs could effectively bridge the gap between cognitive biases and safe behaviors.

In high-altitude construction work, where tasks are often complex and time-sensitive, workers frequently rely on the automatic system for decision-making. However, experienced workers, through repeated exposure to hazardous situations and the consequences of unsafe behaviors, develop a more refined reflective system. This enhanced reflective system allows them to better recognize and counteract cognitive biases, leading to safer behaviors.

For high-altitude workers, especially those in the construction industry, experience provides a critical buffer against the negative effects of safety cognition bias. Experienced workers are more likely to have encountered a wide range of hazardous situations, allowing them to develop a more accurate internal model of risk assessment. They also learn from past mistakes and observe the consequences of unsafe practices, leading to a stronger adherence to safety protocols and a more conservative approach to risk-taking. In contrast, less experienced workers, lacking this depth of knowledge, are more likely to rely on intuitive judgments that are susceptible to cognitive biases.

Given these mechanisms, we propose that work experience acts as a crucial moderator in the relationship between safety cognition bias and unsafe behaviors. Specifically, we hypothesize that the impact of safety cognition bias on unsafe behaviors will be weaker among workers with more extensive experience compared to those with limited experience. This moderating effect is expected to be more pronounced among the new generation of high-altitude workers, who may lack the depth of experience needed to counteract cognitive biases effectively.

Therefore, we propose the following hypothesis:

H2. 

Work experience plays an important role in moderating the relationship between safety cognition bias and unsafe behaviors, and this moderating effect varies among high-altitude workers of different age groups.

2.4. The Mediating Role of Risk-Taking Tendencies

Risk-taking propensity refers to an individual’s willingness to engage in behaviors that involve potential risks or uncertainties, with the expectation of achieving a desired outcome or reward. In the context of high-altitude work, this can manifest as a tendency to bypass safety protocols, take shortcuts, or perform tasks in a manner that deviates from established safety standards. Risk-taking propensity is influenced by factors such as personal attitudes toward risk, perceived benefits of risky behaviors, and the perceived likelihood of negative consequences [35]. For example, a worker with a high risk-taking propensity might choose to work faster to meet deadlines, even if it means skipping safety checks or not using proper equipment.

Existing research has identified risk-taking propensity as a critical mediator between safety cognition bias and unsafe behaviors. For instance, studies have shown that cognitive biases can lead to increased risk-taking tendencies, which, in turn, result in higher likelihood of unsafe behaviors. This relationship is further influenced by generational differences, with younger workers exhibiting higher risk-taking propensity due to their pursuit of innovation and experience accumulation. In contrast, older workers, with their conservative approach and adherence to established norms, display lower risk-taking tendencies.

In the field of high-altitude operations, there are significant differences in the cognition and concepts of safety issues among practitioners of different age groups [36]. Specifically, the new generation of high-altitude workers tends to overestimate their own abilities and underestimate the risks of accidents, thus being more willing to explore and adopt new methods or technologies, showing a higher willingness to take risks. In contrast, although the older generation of high-altitude workers are in an environment of rapid change and high work pressure in the construction industry, they rely on their rich work experience and keen insight into safety issues and are more inclined to adopt a prudent work strategy to ensure the stability of work quality and safety [37,38]. Based on the above analysis, we propose the following hypothesis:

H3. 

Safety cognition bias has a positive impact on individuals’ risk-taking propensity, and this impact varies significantly among high-altitude workers of different age groups.

In the construction industry, high-altitude workers often find themselves in potentially dangerous working environments, where they may be inclined to take risky actions to improve work efficiency [39]. However, such risky actions often induce unsafe behaviors, thereby increasing the risk of safety accidents. Some scholars have observed a close correlation between the risk-taking propensity and unsafe behaviors of construction workers, implying that their tendency to take risks may lead them to make unsafe choices during construction operations, thus increasing the likelihood of unsafe actions [40]. Further exploration of the relationship between risk-taking propensity and behavior reveals that individuals with a high propensity for risk-taking often lack an accurate perception of the adventurous nature of their actions, which easily prompts them to engage in risky behaviors [27,41]. It is worth noting that there are significant differences in the degree of risk-taking propensity among high-altitude workers of different age groups. Based on this, we propose the following hypothesis:

H4. 

Risk-taking propensity has a positive impact on unsafe behaviors, and this impact varies among high-altitude workers of different age groups.

Existing research has identified risk-taking propensity as a critical mediator between safety cognition bias and unsafe behaviors [42]. For instance, studies have shown that cognitive biases can lead to increased risk-taking tendencies, which, in turn, result in higher likelihood of unsafe behaviors. This relationship is further influenced by generational differences, with younger workers exhibiting higher risk-taking propensity due to their pursuit of innovation and experience accumulation [43]. In contrast, older workers, with their conservative approach and adherence to established norms, display lower risk-taking tendencies [44,45]. This generational difference in risk-taking propensity underscores its mediating role in the relationship between safety cognition bias and unsafe behaviors. Based on this, we propose the following hypothesis.

To better understand the mediating role of risk-taking propensity, we refer to the “Theory of Planned Behavior (TPB)”, which provides a comprehensive framework for understanding how cognitive biases influence risk-taking propensity and subsequent unsafe behaviors. According to TPB, behavior is influenced by behavioral intention, which is determined by three key factors: attitude toward the behavior, subjective norms, and perceived behavioral control.

Cognitive biases can shape workers’ attitudes and perceived behavioral control, leading to higher risk-taking propensity. For example, younger workers may have a more positive attitude toward risk-taking due to their higher tolerance for uncertainty and lower perceived risk. This increased risk-taking propensity, in turn, leads to a higher likelihood of engaging in unsafe behaviors. In contrast, older workers, with their more conservative attitudes and stronger adherence to safety norms, exhibit lower risk-taking propensity. This generational difference in risk-taking propensity underscores its mediating role in the relationship between safety cognition bias and unsafe behaviors.

Given this theoretical framework, we propose the following hypothesis:

H5. 

Risk-taking propensity plays an intermediary bridging role between safety cognition bias and unsafe behaviors, and this mediating effect varies among high-altitude workers of different age groups.

In summary, the hypothetical model between safety cognitive bias, risk-taking tendencies, work experience, and unsafe behavior is shown in Figure 1.

3. Research Methodology

While previous studies have typically relied on simpler statistical methods or qualitative analyses, this study utilizes Structural Equation Modeling (SEM) to analyze the complex relationship between safety cognitive biases, risk-taking propensities, work experience, and unsafe behaviors. Compared to traditional regression methods, structural equation modeling allows for the simultaneous estimation of multiple relationships and the incorporation of latent variables, thus providing a more robust and comprehensive analysis. Previous studies have often overlooked the moderating role of work experience in the relationship between cognitive bias and unsafe behavior. The present study explicitly explored how work experience affects this relationship, particularly across different age groups. By using work experience as a moderator, we can better understand how knowledge accumulation and situational awareness influence safety behavior.

To ensure the reliability and validity of our findings, this study utilized robustness tests such as the Bootstrap method, outlier detection, and model simplification/expansion. These tests helped to verify that our results were not affected by sample selection, model assumptions, or other potential biases. Previous studies have often lacked this level of validation, which makes our results more plausible and generalizable.

3.1. Scale Design

During the scale design process, to ensure the reliability and validity of the measurement tool, it is common to draw on well-validated mature scales [46]. This study involves four core variables: Safety Cognition Bias (SCB), Unsafe Behavior (UB), Work Experience (WE), and Risk-Taking Tendency (RT). The mature scales selected are closely related to these research variables, and their reliability and validity have been strictly verified [47,48]. Based on the match of each scale in terms of the number of items, form, and content with the research needs, and in combination with expert consultation opinions, we made appropriate adjustments to the scales and finally adopted the Likert 5-point rating system. Specifically:

(1)

The compilation of the Safety Cognition Bias (SCB) scale mainly refers to the safety cognition bias influencing factors scale, aiming to assess the deviations in individuals’ cognition, understanding, and judgment of safety issues, and it includes five items [49].

(2)

The Unsafe Behavior (UB) scale is compiled based on the unsafe behavior scale, covering four aspects of items such as non-compliance with safety and non-participation in safety, to comprehensively reflect the unsafe behavior performance of individuals [50].

(3)

The Work Experience (WE) scale is formulated by referring to the work experience scale, involving four key items such as job age and experience relevance, to evaluate the level of individuals’ work experience [51].

(4)

The Risk-Taking Tendency (RT) scale draws on the accident proneness scale, including five items such as risk-taking tendency and self-control, to measure the tendency of individuals’ risky behaviors [52].

After several rounds of surveys, consultations, and validations, the final scale has shown excellent reliability and validity, meeting the requirements for formal measurement.

During the scale design process, in order to ensure the reliability and validity of the measurement tool, we referred to proven scales and made appropriate adjustments in accordance with the research needs. Specifically, for the measurement of Work Experience (WE), we referred to existing work experience scales and developed new entries based on the characteristics of high-altitude construction workers. These entries included Years of Service, Relevant Experience, Training Experience, and Accident Experience. Each entry was rated on a 5-point Likert scale from 1 (very little/never) to 5 (very much/always). By calculating the total score for these four entries, we categorized work experience into three levels: Low Experience (total score below 10), Moderate Experience (total score between 10 and 15), and High Experience (total score above 15). This categorization takes into account not only years of experience but also factors such as training and accident experience and provides a more comprehensive picture of a worker’s level of experience.

3.2. Data Sources

This study focuses on the survey of high-altitude workers in the construction industry, spanning from 29 February 2024 to 30 April 2024. During this period, we distributed anonymous questionnaires to high-altitude workers in several construction-related enterprises in China. The questionnaire design included three screening criteria: “type of work unit”, “job position”, and “high-altitude work certificate”, aiming to accurately target the research group. After the survey, we collected a total of 371 questionnaire responses, of which 311 were valid, with a high valid return rate of 83.83%, meeting the basic requirements of sample size in statistics.

To minimize geographical bias, the sample was drawn from multiple regions across China, including both urban and rural areas. To reduce industry bias, the survey included workers from various types of construction projects, such as residential, commercial, and infrastructure development. To address self-report bias, the questionnaire was designed to be anonymous, and responses were cross-validated with on-site observations and accident records where possible. These measures were taken to ensure the representativeness and reliability of the data.

In-depth analysis of the valid samples revealed that male employees accounted for as high as 86.37%, while female employees were 13.63% (

Table 1. Descriptive statistics of respondents’ characteristics.

Statistic	Form	Frequency	Percentage (%)
Age	Born after 1980 (aged 44 and below)	203	65.39%
	Born before 1980 (44+)	108	34.61%
Genders	Male	269	86.37%
	Female	42	13.63%
Length of service	Less than 5 years	129	41.51%
	5 to 10 years	149	48.05%
	More than 10 years	32	10.44%
Educational attainment	High school and below	127	40.90%
	College and above	184	59.10%

). In terms of educational background, employees with a high school diploma or lower education accounted for 40.90%, and those with an associate degree or higher accounted for 59.10%. Regarding age structure, employees born in 1980 or later (i.e., 44 years old and under) accounted for 65.39%, while those born before 1980 (i.e., over 44 years old) accounted for 34.61%. In terms of work experience, employees with less than 5 years of experience accounted for 41.51%, those with 5 to 10 years of experience accounted for 48.05%, and those with more than 10 years of experience accounted for 10.44%. This distribution well reflects the overall characteristics of high-altitude workers in the construction industry.

To further explore the differences between different age groups, this study, based on the age classification standards of the World Health Organization (WHO), divided all survey subjects into the new generation group (44 years old and under) and the older generation group (over 44 years old), with January 1 1980 as the dividing line (with 2024 as the reference year). At the same time, all survey subjects together constituted the overall group.

3.3. Reliability Test and Correlation Analysis

We utilized SPSS software (Version: IBM SPSS Statistics 20) to conduct reliability and validity tests on the collected survey data [53]. The results showed that the KMO value of the data was as high as 0.941, and the Sig. value of Bartlett’s test of sphericity was 0.00, both meeting the prerequisites for factor analysis. In the overall group, the Cronbach’s alpha coefficients for the observed variables-safety cognition bias, risk-taking propensity, work experience, and unsafe behavior-were 0.874, 0.891, 0.789, and 0.901, respectively. The composite reliability (CR) values were 0.853, 0.879, 0.837, and 0.868, all exceeding the threshold of 0.7. Meanwhile, their average variance extracted (AVE) values were 0.567, 0.626, 0.551, and 0.552, all greater than the standard of 0.5. These data fully demonstrate that the scales we used have good internal consistency and convergent validity.

Factor analysis was performed to assess the underlying structure of the measured variables and to ensure that the items in each scale represent distinct constructs. The principal component analysis with varimax rotation was used to extract factors. The results indicated that the items loaded well onto their respective factors, with factor loadings ranging from 0.65 to 0.89 for all scales, exceeding the commonly accepted threshold of 0.50. The extracted factors explained 72.3% of the total variance, demonstrating that the scales captured the essential dimensions of the constructs.

Specifically, for the Safety Cognition Bias (SCB) scale, three factors were extracted, explaining 68.5% of the variance, with factor loadings ranging from 0.70 to 0.85. For the Risk-Taking Tendency (RT) scale, two factors were identified, accounting for 75.2% of the variance, with loadings between 0.65 and 0.80. The Work Experience (WE) scale resulted in a single factor that explained 65.4% of the variance, with loadings from 0.72 to 0.82. Lastly, the Unsafe Behavior (UB) scale yielded two factors, capturing 70.1% of the variance, with loadings from 0.75 to 0.89. These results confirm that the scales have a clear factor structure and are suitable for further analysis.

In the overall group, the Cronbach’s alpha coefficients for the observed variables-safety cognition bias, risk-taking propensity, work experience, and unsafe behavior-were 0.874, 0.891, 0.789, and 0.901, respectively. The composite reliability (CR) values were 0.853, 0.879, 0.837, and 0.868, all exceeding the threshold of 0.7. Meanwhile, their average variance extracted (AVE) values were 0.567, 0.626, 0.551, and 0.552, all greater than the standard of 0.5. These data fully demonstrate that the scales we used have good internal consistency and convergent validity.

Next, we analyzed the correlations between variables, with the specific results shown in Figure 2.

From the figure, the following can be observed.

In the overall group, there is a significant positive correlation between safety cognition bias and unsafe behavior (correlation coefficient r = 0.719, p < 0.01), and a significant positive correlation also exists between safety cognition bias and risk-taking propensity (r = 0.663, p < 0.01). Moreover, a significant positive correlation is observed between risk-taking propensity and unsafe behavior (r = 0.852, p < 0.01).

In the older generation group, although the strength of these correlations varies, they remain significant. The correlation coefficient between safety cognition bias and unsafe behavior is 0.643 (p < 0.01), between safety cognition bias and risk-taking propensity is 0.693 (p < 0.01), and between risk-taking propensity and unsafe behavior reaches 0.892 (p < 0.01).

In the new generation group, these correlations are still significantly present. Specifically, the correlation coefficient between safety cognition bias and unsafe behavior is 0.749 (p < 0.01), between safety cognition bias and risk-taking propensity is 0.677 (p < 0.01), and between risk-taking propensity and unsafe behavior is as high as 0.831 (p < 0.01).

The differences in the correlation coefficient r reveal the differences among construction workers across different generations, further indicating that generational differences may have an impact on the path relationships between variables. The above findings are consistent with our theoretical expectations.

4. Model Validation and Analysis

4.1. Model Fit

Utilizing AMOS software (Version: IBM SPSS Amos 27), we constructed structural equation models for three different groups, with the structural equation model for the overall group presented in Figure 3. We conducted fit-index tests for these models and summarized the results in

Table 2. Results of goodness-of-fit tests for each model.

Fitness Index	${\mathsf{\chi }}^2$	${\mathsf{\chi }}^2/\mathbf{d}\mathbf{f}$	GFI	AGFI	RMSEA	NNFI	IFI	CFI
Model M (overall group)	185.321	2.415	0.927	0.897	0.074	0.951	0.962	0.963
Model Ma (old generation group)	71.177	1.077	0.907	0.847	0.034	0.924	0.966	0.968
Model Mb (Cenozoic group)	141.147	1.948	0.909	0.871	0.071	0.925	0.966	0.967
Fitting result	eligible	eligible	eligible	eligible	eligible	eligible	eligible	eligible

. According to the data presented in Table 2, all key fit indices of the models reached the generally accepted standard range. Specifically, the ratio of Chi-square to degrees of freedom (Chi-square/df) was less than 3, and the goodness-of-fit index (GFI), non-normed fit index (NNFI), incremental fit index (IFI), and comparative fit index (CFI) all exceeded 0.9, with the adjusted goodness-of-fit index (AGFI) being greater than 0.8, and the root mean square error of approximation (RMSEA) being below 0.08. These key fit indices met the established acceptable standards, indicating that all models had good fit.

Figure 3 presents the structural equation model (SEM) for the overall group, illustrating the relationships among safety cognition bias (SCB), risk-taking propensity (RT), unsafe behavior (UB), and the moderating role of work experience (WE). The model was validated using AMOS software, and the fit indices (Table 2) confirm that the model fits the data well.

In Figure 3, SCB (Safety Cognition Bias) is shown to have a significant positive direct effect on UB (Unsafe Behavior) (β = 0.221, p < 0.001), indicating that higher levels of safety cognition bias directly lead to increased unsafe behaviors among high-altitude workers. This direct effect is stronger in the new generation group (β = 0.297, p < 0.001) compared to the older generation group (β = 0.050, p > 0.05), suggesting that younger workers are more susceptible to the negative impact of cognitive biases on their behavior.

SCB also significantly influences RT (Risk-Taking Propensity) (β = 0.767, p < 0.001), which, in turn, affects UB (β = 0.791, p < 0.001). This indicates that risk-taking propensity partially mediates the relationship between safety cognition bias and unsafe behavior. Specifically, in the new generation group, RT plays a partial mediating role (indirect effect: β = 0.557, p < 0.001), while in the older generation group, RT fully mediates this relationship (indirect effect: β = 0.611, p < 0.001). This suggests that older workers’ unsafe behaviors are primarily driven by their risk-taking propensity rather than direct cognitive biases.

Table 3. Comparison of standardized path coefficients and significance across models.

Trails	Standardized Path Factor			Comparison of Significance	Conclude
	M (Overall Group)	Ma (Old Generation Group)	Mb (Cenozoic Group)
H1 (Safety perception bias → unsafe behavior)	0.221 ***	0.051 (0.472)	0.297 ***	Cenozoic group > old generation group	Support
H3 (Safety perception bias → risk-taking tendency)	0.767 ***	0721 ***	0.786 ***	Remarkable	Support
H4 (Risk-taking tendencies → unsafe behavior)	0.791 ***	0.942 ***	0.711 ***	Remarkable	Support

Note: ***, p < 0.001.

presents the standardized + ath coefficients and their significance comparison results for each model. Analyzing the data in Table 3, it can be observed that in the overall sample group, all standardized path coefficients of the paths are positive, and the corresponding p-values are significantly below the 0.001 level, clearly indicating that these paths are statistically significant. Further comparison of the data between the older and new generations shows differences in the standardized coefficients and p-values, reflecting the varying path influence strengths across different generations. Therefore, it can be confirmed that the path relationships described in Hypotheses H1, H3, and H4 are empirically supported.

4.2. Moderating Effects Test

Hierarchical regression analysis was employed to test the moderating role of work experience [54,55], with the results presented in

Table 4. Regression analysis of the moderating effect of work experience.

Variant	Implicit Variable	M (Overall Group)			Ma (Old Generation Group)			Mb (Cenozoic Group)
		Model 1	Model 2	Model 3	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3
Control variable	Genders	0.071	0.036	0.030	0.023	0.167	0.161	0.038	0.000	−0.004
	Job category	0.004	−0.019	−0.021	0.081	−0.052	−0.057	−0.022	−0.003	−0.007
	Length of service	0.072	0.037	0.039	−0.028	−0.039	−0.040	0.111	0.067	0.070
	Educational attainment	0.123 *	0.018	0.027	−0.031	−0.114	−0.108	0.148	0.021	0.034
Independent variable	Security perception bias		0.704 ***	0.651 ***		0.668 ***	0.717 ***		0.741 ***	0.651 ***
Moderator variable	Working experience		0.027	0.021		−0.179 *	−0.181 *		0.098	0.091
Interaction term	Safety perception bias × work experience			0.121 *			−0.078			0.237 ***
Model statistic	R2	0.031	0.511	0.517	0.059	0.491	0.497	0.039	0.561	0.593
	∆R2	0.031	0.480	0.006	0.059	0.432	0.006	0.039	0.522	0.032
	F	1.960	141.348 ***	5.881 *	1.258	34.771 ***	0.657	1.747	116.868 ***	15.917 ***

Note: * is, p < 0.05; *** indicates, p < 0.001. “Safety perception bias × work experience” means the interaction term, that is, the interaction between safety cognitive bias and work experience. The “×” here is not a simple multiplication operation, but rather represents the interaction effect between two variables.

.

Based on the data analysis in Table 4, the following conclusions can be drawn: In the overall group sample, the interaction between safety cognition bias and work experience has a significant positive effect on unsafe behavior, with a standardized path coefficient β of 0.121, and a p-value less than 0.05, indicating that this relationship is statistically significant.

For the older generation of high-altitude workers, the interaction between safety cognition bias and work experience does not significantly affect their unsafe behavior, with a β value of -0.078 and a p-value greater than 0.05, meaning that in this group, the interaction does not show obvious statistical significance.

For the new generation of high-altitude workers, the interaction between safety cognition bias and work experience has a significant positive impact on their unsafe behavior, with a β value reaching 0.237 and a p-value less than 0.001, highly significant. This indicates that in the new generation group, as work experience increases, the positive effect of safety cognition bias on unsafe behavior will significantly decrease.

In summary, the moderating role of work experience varies across different generations, thereby validating the correctness of Hypothesis H2.

To visually illustrate the moderating effect of work experience, we have created a moderation effect diagram (see Figure 4). The figure reveals an important trend: for high-altitude workers with little work experience, the positive driving effect of safety cognition bias on unsafe behavior is relatively pronounced; however, as work experience accumulates, this positive effect gradually weakens. Particularly noteworthy is that among the new generation of high-altitude workers, the phenomenon of this effect being weakened by work experience is more pronounced.

4.3. Mediation Effect Test

The mediating effect of risk-taking propensity was tested using the bias-corrected method [56], with the results detailed in

Table 5. Tests for the mediating effect of propensity to take risks.

Trails	Modelling	Typology	Efficiency Value	Standard Error	95% Confidence Intervals			Percentage of Effect/%	Result
					Upper	Lower Limits	p
Safety perception bias → unsafe behaviour	M (overall group)	Direct effect	0.211	0.070	0.347	0.078	0.006	27.47	Partial mediation
		Indirect effect	0.557	0.079	0.701	0.411	0.001	72.53
		Total effect	0.768	0.067	0.872	0.639	0.001
	Ma (old generation group)	Direct effect	0.050	0.141	0.221	−0.261	0.670	7.56	Partial mediation
		Indirect effect	0.611	0.171	0.981	0.401	0.001	92.44
		Total effect	0.661	0.114	0.870	0.437	0.001
	Mb (Cenozoic group)	Direct effect	0.284	0.101	0.491	0.108	0.007	35.15	Full mediation
		Indirect effect	0.524	0.089	0.711	0.321	0.001	64.85
		Total effect	0.808	0.071	0.907	0.647	0.001

. The mediating effect was further divided into direct and indirect effects to thoroughly analyze the specific role of the mediating variable in this relationship [57,58].

According to the data analysis in Table 5, it was found that in the overall group (Model M) and the new generation group (Model Mb), the confidence intervals (CI) of both direct and indirect effects did not include the value 0, confirming that risk-taking propensity plays a partial mediating role between safety cognition bias and unsafe behavior. Similarly, in the older generation group (Model Ma), the CI value of the direct effect included 0, while the CI value of the indirect effect did not include 0, showing a suppression effect [59]; that is, the direct effect does not exist, indicating that risk-taking propensity plays a full mediating role between safety cognition bias and unsafe behavior. It is worth noting that both partial and full mediation effects belong to the category of mediating effects, thereby validating the correctness of Hypothesis H5.

The existence of partial mediating effects means that the unsafe behavior of the new generation of high-altitude workers is not only affected by safety cognition bias but also closely related to their inherent risk-taking propensity. Due to their relatively limited experience and the need for enhanced safety awareness, the new generation of high-altitude workers may unconsciously take unsafe actions when performing tasks, thereby increasing the risk of accidents. Moreover, they tend to ignore safety regulations and procedures, are more willing to take risks, and turn a blind eye to potential dangers, thus exacerbating the occurrence of unsafe behaviors. In contrast, in the older generation of high-altitude workers, the existence of full mediating effects indicates that pure safety cognition bias is not sufficient to directly lead to unsafe behavior. Because the older generation of high-altitude workers have long working years and rich experience, their perception and assessment ability of safety risks are relatively strong, so the impact of safety cognition bias on their behavior is relatively limited. However, once the older generation of high-altitude workers have a high propensity for risk-taking, they may disregard safety regulations and thus trigger unsafe behaviors.

4.4. Robustness Test

In order to further validate the robustness of the findings, a variety of methods were used to test the robustness of the model to ensure that the findings were not affected by sample selection, model assumptions, or other potential factors.

1.

Bootstrap method

We re-estimated the sample data using the Bootstrap method (repeated sampling 1000 times) to verify the stability of the path coefficients. The results show that the 95% confidence intervals of all critical path coefficients do not contain zero values, indicating that these paths maintain significance in repeated sampling. For example, the path coefficients of the direct effect of safety perception bias on unsafe behaviors in the overall sample remain significant in the Bootstrap test (with confidence intervals ranging from 0.187 to 0.235), which further validates the robustness of the model results.

2.

Outlier detection and exclusion

We performed outlier detection on the sample data and reran the model after excluding possible outliers. By comparing the results of the model with outliers included and excluded, it was found that the magnitude of change in the critical path coefficients was within acceptable limits (e.g., the change in the path coefficients of the safety perception bias for unsafe behaviors was less than 10%). This indicates that the results of the study are robust to outliers.

3.

Model simplification and extension

We conducted the robustness test by simplifying and extending the model. In the simplified model, some control variables (e.g., education level and job category) were removed and the model was re-run; in the extended model, additional control variables (e.g., workplace safety culture) were added. The results show that the critical path coefficients remain consistent in both the simplified and extended models, further validating the robustness of the model results.

5. Results and Discussion

5.1. Results

The results of the study revealed significant differences in safety perception bias and risk-taking behavior between younger and older workers. Younger workers exhibited higher levels of safety perception bias (mean = 3.5; standard deviation = 0.6), suggesting that they are more likely to underestimate safety risks and overestimate their personal capabilities, thereby ignoring the importance of safety measures. In contrast, older workers had a relatively lower level of safety perception bias (mean = 2.8; standard deviation = 0.5), suggesting that they are more accurate in assessing safety risks and more inclined to comply with safety norms. In terms of risk-taking behavior, younger workers showed a higher propensity for risk-taking behavior (mean = 4.0; standard deviation = 0.7) and were more willing to adopt riskier behaviors on the job, such as skipping safety checks or not wearing personal protective equipment. Older workers, on the other hand, had a lower propensity for risky behavior (mean = 2.9; standard deviation = 0.6) and were more inclined to adopt a cautious attitude and strictly follow safety norms on the job.

Further mediation effect analysis showed that risk-taking behavior mediated between safety perception bias and unsafe behavior. For younger workers, risk-taking behavior played a partial mediating role (indirect effect share = 72.53%), suggesting that risk-taking behavior is an important mediator of unsafe behavior among younger workers. For older workers, on the other hand, risk-taking behavior played a fully mediating role (proportion of indirect effect = 92.44%), indicating that older workers’ unsafe behaviors were mainly manifested through risk-taking behavior.

Work experience moderated the relationship between safety perception bias and unsafe behaviors, which was especially significant among younger workers. The positive effect of safety perception bias on unsafe behaviors was significantly attenuated in younger workers with increasing work experience (β = 0.237, p < 0.001), suggesting that work experience is effective in mitigating the effect of safety perception bias on unsafe behaviors. In contrast, the effect of work experience on older workers was not significant (β= −0.078, p > 0.05), suggesting that older workers have accumulated enough experience in long-term work to better cope with safety cognitive bias.

This study reveals that safety cognition bias significantly impacts unsafe behaviors among high-altitude workers, with a stronger effect observed in the new generation compared to the older generation. Risk-taking propensity is found to mediate this relationship, with partial mediation for the new generation and full mediation for the older generation. Work experience significantly moderates the relationship between safety cognition bias and unsafe behaviors, particularly among the new generation. These findings highlight the critical role of generational differences in shaping safety behaviors.

There are some similarities and differences between the results of this study and previous studies. For example, related studies have shown that younger workers exhibit higher levels of safety perception bias, which is consistent with the results of this study. However, there are also studies that found no significant difference in the effect of work experience on safety perception bias between different age groups, which is different from the results of the present study in which the effect of work experience is significant for younger workers but not for older workers. In addition, this study analyzed the mediating role of risk-taking behavior through structural equation modeling (SEM) and found it to be partially mediated among younger workers and fully mediated among older workers, a finding that provides a new perspective for understanding differences in worker behavior across age groups. These results suggest that although there are general differences in safety cognitive biases between younger and older workers, their behavioral performance and regulatory mechanisms may vary depending on work experience and risk-taking tendencies.

Work experience plays a crucial role in moderating the relationship between safety cognition bias and unsafe behavior. Specifically, work experience among the new generation of high-altitude workers exerts a significant positive moderating effect, whereas no such moderating effect is observed among the older generation. Given this finding, establishing a mechanism for the exchange of work experience between new and experienced high-altitude workers is essential. This approach facilitates the accumulation, summarization, and transfer of work experience, thereby enhancing high-altitude work skills and reducing the risk of accidents.

Risk-taking propensity significantly predicts unsafe behavior among high-altitude workers and acts as a mediator between safety cognition bias and unsafe behavior. Specifically, among the new generation of high-altitude workers, risk-taking propensity plays a partial mediating role, while it serves as a full mediator among the older generation. Therefore, the implementation of strict safety regulations and enhanced supervision and training are critical. These measures can help high-altitude workers better understand and comply with safety protocols, thereby effectively reducing the impact of risk-taking propensity on unsafe behavior.

5.2. Discussion

5.2.1. Discussion on Impact and Association

The findings of this study underscore the importance of addressing generational differences in safety management strategies within the construction industry. For the new generation of high-altitude workers, targeted safety training programs are essential to mitigate the impact of safety cognition bias and reduce risk-taking behaviors. These programs should focus on enhancing situational awareness and reinforcing safety protocols. For the older generation, leveraging their extensive experience through mentorship programs could help transfer critical safety knowledge to younger workers while maintaining their adherence to safety regulations.

While this study reveals the impact of safety perception bias and risk propensity on unsafe behaviors of workers working at height, safety training programs need to be further customized for the specific challenges of the construction industry. For young workers, training should incorporate actual risks on construction sites, such as falls from height and equipment operation, and reinforce their safety awareness and operational skills through simulated work-at-height scenarios and emergency drills. At the same time, immersive training using digital tools (e.g., VR and AR) can help them quickly adapt to the complex working environment. For older workers, training should focus on the application of new technologies and updating of safety standards, such as the use of intelligent safety equipment and the latest fall protection technologies. In addition, taking into account the characteristics of the construction industry, cross-generational training modules should be developed to allow experienced older workers to share their practical experience in working at height, while allowing younger workers to enhance their safety efficiency through technological means. Through customized training programs, the construction industry can effectively reduce the risk of work-at-height accidents and improve overall safety management effectiveness.

The significant differences observed between the new and older generations in safety cognition bias and unsafe behaviors can be attributed to multiple factors. First, the new generation of workers, with limited work experience, tends to rely more on intuitive judgments, which are susceptible to cognitive biases [30]. In contrast, older workers, with their accumulated experience, are better equipped to engage the reflective system and counteract these biases [60]. Second, the sociocultural environment in which the new generation grew up emphasizes innovation and efficiency, which may lead to a higher propensity for risk-taking. In contrast, the older generation, having experienced the consequences of unsafe behaviors, adopts a more conservative approach to safety [61]. Lastly, the rapid technological advancements in recent years have influenced the new generation’s perception of safety risks, making them more likely to underestimate potential dangers.

For the younger generation of workers, training should focus on enhancing their situational awareness and risk perception, for example, through virtual reality (VR) simulations and real-life scenario drills, to help them identify potential hazards and reduce their propensity to take risks. At the same time, a mentor system should be set up to allow experienced older workers to pass on their safety knowledge and make up for the lack of experience of younger workers. For older workers, training should reinforce existing safety knowledge, update safety standards, and introduce new technologies to enhance safety management. In addition, regular feedback and evaluation of the effectiveness of the training should be carried out to ensure that the training content is closely aligned with the actual needs. Through these measures, safety training programs can not only effectively reduce unsafe behaviors of workers working at height but also promote the exchange of experience between workers of different generations and enhance the overall safety management level of the construction industry.

Moreover, the significant moderating role of work experience highlights the need for continuous skill development and knowledge sharing among workers. Establishing mechanisms for intergenerational collaboration can bridge the gap between experience and new technologies, ensuring a safer working environment for all. The full mediation effect of risk-taking propensity among the older generation suggests that maintaining a conservative approach to safety is crucial, even with years of experience.

In summary, this study provides valuable insights into the complex interplay between safety cognition bias, risk-taking propensity, and unsafe behaviors across different age groups. By addressing these factors through tailored interventions, the construction industry can significantly enhance safety performance and reduce the incidence of accidents. However, the generalizability of these findings is limited by the specific context of the Chinese construction industry. While the commonalities of work-at-height risks, the importance of safety norms, and the impact of worker behaviors on accidents are widespread in the construction industry, differences in culture, laws, and management practices may affect the applicability of the results. Construction workers in different countries may exhibit varying cognitive and behavioral responses to safety risks due to differences in safety training and cultural values. Additionally, stricter safety regulations and better safety management systems may reduce the influence of safety perception bias on unsafe behaviors. Therefore, although the results of this study have important implications for the Chinese construction industry, their generalizability needs to be validated in a broader context. Future studies could conduct similar investigations in different countries or regions to validate the cross-cultural applicability of the results and to explore how cultural, legal, and managerial differences affect the relationship between safety perception bias and unsafe behaviors.

While this study provides valuable insights into the impact of safety cognition bias and risk-taking propensity on unsafe behaviors among high-altitude workers, it is important to acknowledge the potential influence of confounding factors such as job role changes and organizational safety culture. In the dynamic environment of the construction industry, workers often experience job role changes, which can introduce new safety risks and challenges. For example, transitioning from a ground-based role to high-altitude work may require workers to adapt to different safety protocols and hazards. This adaptation period can be vulnerable to cognitive biases, as workers may underestimate the risks associated with their new roles. Additionally, role changes can affect workers’ psychological states, potentially increasing their risk-taking propensity. Future research should explore how job role changes influence safety behaviors and cognitive biases and how targeted interventions can support workers during these transitions.

The safety culture within an organization plays a crucial role in shaping workers’ attitudes and behaviors towards safety. A strong safety culture can mitigate the impact of cognitive biases and promote adherence to safety protocols, while a weak safety culture may exacerbate unsafe behaviors. This study did not account for the varying safety cultures of the organizations involved, which may limit the generalizability of the findings. Future research should incorporate organizational safety culture as a key variable to better understand its interaction with cognitive biases and risk-taking tendencies. This can help identify strategies to foster a positive safety culture that supports workers at all levels of experience.

In conclusion, while this study highlights the importance of addressing generational differences and cognitive biases in safety management, the influence of job role changes and organizational safety culture cannot be overlooked. Future research should consider these confounding factors to provide a more comprehensive understanding of the determinants of safety behaviors in the construction industry. By doing so, we can develop more effective safety interventions that account for the complex interplay of cognitive, organizational, and contextual factors.

5.2.2. Policy Implications and Recommendations

To enhance the safety of work-at-height operations in the construction industry, it is recommended that mandatory safety training programs be developed, incorporating virtual reality (VR) and augmented reality (AR) technologies to provide customized training content for different generations of workers, and at the same time, to promote cross-generational mentorship programs to facilitate experience transfer to implement standardized reporting and safety audits to identify best practices and to continuously evaluate the effectiveness of training to ensure its long-term validity. These measures will help reduce accidents and improve the overall safety level of the industry.

(1)

Mandatory Tailored Safety Training Programs

Regulatory bodies and industry associations should consider mandating safety training programs that are tailored to the specific needs of different generational groups. For younger workers, these programs could incorporate advanced technologies such as virtual reality (VR) and augmented reality (AR) to simulate real-life scenarios and improve situational awareness. For older workers, training could focus on updating their knowledge of new safety technologies and reinforcing existing safety protocols.

(2)

Intergenerational Mentorship Programs

Policies should encourage the development of mentorship programs within construction companies to facilitate knowledge transfer from experienced older workers to younger ones. These programs could help mitigate cognitive biases and reduce unsafe behaviors by leveraging the extensive experience of older workers.

(3)

Standardized Reporting and Safety Audits

Policies should promote the documentation and sharing of work experience through standardized reporting systems and safety audits. This could help identify best practices and areas for improvement across different construction sites, thereby reducing the incidence of accidents.

(4)

Continuous Evaluation of Safety Training Programs

Regulatory frameworks should emphasize the continuous evaluation and updating of safety training programs based on empirical evidence. This could involve periodic assessments of training effectiveness and the incorporation of feedback from workers to ensure that safety interventions remain relevant and effective over time.

By implementing these policy recommendations, the construction industry can create a safer working environment for high-altitude workers and reduce the incidence of accidents. These interventions are crucial for addressing the generational differences in safety cognition and behavior and for leveraging the experience of older workers to enhance the safety of younger workers.

6. Limitations and Future Prospects

The current study, despite its comprehensive approach, has several limitations. First, the large-scale data collection may have introduced errors due to insufficient representation of certain subgroups, which could affect the precision of capturing complex relationships and subtle differences between variables. Second, the study’s time span limited the ability to track long-term dynamic changes in safety cognition bias, work experience accumulation, risk-taking propensity, and the formation of unsafe behaviors among high-altitude workers. These limitations highlight the need for more detailed and extended research to fully understand the underlying mechanisms influencing safety behaviors in the construction industry.

One potential limitation is the possibility of biases in the sample, including geographical, industry, and self-report biases. To address these concerns, we ensured that the sample was drawn from multiple regions and included workers from various types of construction projects. Additionally, we used anonymous questionnaires and cross-validated self-reported data with on-site observations and accident records to minimize self-report bias. Future research should further explore these biases and their impact on safety behaviors in the construction industry.

Future research should focus on exploring the underlying causes of generational differences in safety cognition and behavior. This includes examining the role of sociocultural backgrounds, educational levels, and technological advancements in shaping these differences. Additionally, longitudinal studies could provide insights into how these factors evolve over time and influence safety behaviors. By addressing these limitations and exploring these areas, future research can contribute to more effective safety management strategies that account for the complex interplay of cognitive, organizational, and contextual factors.

To address these limitations and further advance the understanding of safety behaviors among high-altitude workers, several avenues for future research are proposed. Future studies should expand the sample scope to include a more diverse group of workers, enhancing the representativeness and generalizability of the findings. Longitudinal research over an extended period is recommended to accurately assess the long-term dynamics of safety cognition bias and its impact on behavior. Additionally, integrating advanced technologies such as virtual reality (VR) and artificial intelligence (AI) could provide valuable insights into workers’ safety behavior responses under realistic conditions. Exploring intergenerational knowledge transfer mechanisms and investigating cultural and regional differences in safety behaviors are also suggested as important areas for future research. These directions will not only address the current limitations but also contribute to more effective safety management strategies in the construction industry.