Evaluating the Safety Climate in Construction Projects: A Longitudinal Mixed-Methods Study

Abstract

Safety climate has been extensively studied using survey-based approaches, providing significant insights into safety perceptions and behaviors. However, understanding its dynamics in construction projects requires methods that address temporal and trade-specific variability. This study employs a longitudinal, mixed-methods design to explore safety climate dynamics. Quantitative data analyzed with ANOVA revealed stable overall safety climate scores across project phases, while Item Response Theory (IRT) identified survey items sensitive to safety climate changes. Positive perceptions were associated with management commitment and regular safety meetings, while negative perceptions highlighted challenges such as workplace congestion and impractical safety rules. Qualitative data from semi-structured interviews uncovered trade-specific and phase-specific safety challenges, including issues tied to site logistics and workforce dynamics. For instance, transitioning from structural to interior work introduced congestion-related risks and logistical complexities, underscoring the need for phase-adapted strategies. This combination of quantitative stability and qualitative variability provides empirical evidence of safety climate dynamics in construction. The findings emphasize the importance of tailoring safety interventions to address trade-specific and phase-specific risks. This study advances the understanding of the safety climate in dynamic work environments and offers actionable recommendations for improving construction safety management through targeted, proactive strategies.

1. Introduction

Construction has long been considered one of the most hazardous industries [1]. The high fatality and accident rates are attributed to the complex, decentralized, and dynamic nature of the construction working environment [2]. Over the past few decades, significant efforts have been made to improve construction site safety management, including implementing safety regulations, introducing training programs, and raising awareness of safety practices [3]. Despite these efforts, global construction safety performance has remained at a concerning high plateau over the last decade [4].

Construction safety performance has traditionally been measured using lagging indicators, such as injuries and accidents. However, these indicators are criticized for their reactive and delayed nature [5]. In addition, concerns have been raised about the reliability and reportability of lagging indicators [6]. Consequently, research has increasingly focused on leading indicators [7,8,9], with safety climate emerging as one of the most widely used indicators. Over the past few decades, safety climate has been extensively studied across various trades, including construction [10,11,12]. Meta-analyses have shown a strong correlation between safety climate and safety behavior, establishing safety climate as a diagnostic tool for enhancing safety management [13,14,15].

Safety climate assessments provide valuable insights into workers’ perceptions of safety practices, management commitment, and workplace conditions. These assessments are critical for identifying gaps in safety practices and tailoring interventions to specific contexts. Numerous studies have measured safety climates using various scales to analyze their psychometric properties quantitatively [16,17,18]. However, this approach is challenging in the construction industry for the following reasons: (1) Generic safety climate scales often address global safety concerns and fail to identify site-specific factors, such as trade-specific risks, changing workforce dynamics, and evolving site conditions. This limits their ability to uncover the underlying drivers of safety climate [19]; (2) These methods typically provide a snapshot of a safety climate at a single point in time, failing to reflect the temporal fluctuations inherent in construction projects [20]; (3) The practical implications of safety climate scores are challenging for most construction firms to interpret [21]. For example, a low safety climate score on the safety training scale indicates that workers perceive deficiencies in their safety education but does not specify actionable measures for project management to improve the training’s effectiveness. This creates critical conceptual ambiguity in identifying and addressing the factors that drive a safety climate.

A recent bibliometric analysis by Luo et al. [22] identifies that safety climate remains a central topic in construction safety research, often linked with themes of management and worker behavior. However, there is a need for dynamic and interdisciplinary approaches. By regularly evaluating safety climates, construction managers can foster a positive safety culture, enhance compliance with safety protocols, and monitor the effectiveness of safety programs over time. This dynamic approach ensures that safety strategies remain responsive to the evolving challenges of construction projects, ultimately reducing risks and improving overall safety performance.

Thus, the main objective of this study was to investigate the temporal and trade-specific variability of the safety climate in construction projects. To achieve this, the following questions were proposed:

• How does the safety climate evolve across different phases of construction projects in a dynamic context?
• What trade-specific and contextual factors influence fluctuations in safety climate perceptions?
• How can a longitudinal, mixed-methods approach provide a more nuanced understanding of the dynamics of a safety climate?

The remainder of this paper is structured as follows: Section 2 presents the theoretical background and reviews the literature on safety climates in the construction industry, emphasizing gaps addressed by this study. Section 3 details the research methodology, including the longitudinal mixed-methods approach and data analysis techniques. Section 4 discusses the findings from both quantitative and qualitative analyses, highlighting dynamic changes in safety climate and trade-specific insights. Section 5 discusses the key findings and practical implications of this study. Section 6 outlines limitations and suggests directions for future research. Finally, Section 7 concludes the paper by summarizing key contributions.

2. Background

2.1. Conceptualizing the Safety Climate in Construction

As an organizational construct, the concept of safety climate in the construction industry is fundamentally similar to that in the other sectors. Zohar defines a safety climate as the shared perceptions of employees regarding their work environment [23,24]. Since its introduction to the construction industry in the 1990s [25], extensive studies have been conducted to measure the safety climate [26,27,28] and to establish its causal relationship with safety performance [29,30,31].

Despite the progress in measuring and validating safety climates, the conceptualization of the safety climate in the construction industry remains challenging due to the sector’s inherent complexity and dynamic nature. Most studies on construction safety climates have traditionally focused on the “organization” as the unit of analysis [32]. However, in recent years, safety climate has been suggested as a multilevel concept, including at the organizational and group levels [33,34]. In construction, the notion of an “organization” can be ambiguous. Due to its subcontracting nature, the typical top-down hierarchy structure cannot readily apply to construction projects. On an evolving construction site, multiple subcontractors are loosely connected under the general contractor to work simultaneously and are isolated from their parent companies. For subcontractor workers, it can be unclear whether the “organization” refers to their employer, the general contractor, or the broader project-based organization. The interchangeability of organizational and project safety climates is often taken for granted, and few early construction safety climate studies addressed this conceptual discrepancy [2].

The complexity of the concept of “organization” in construction is not just about fragmented and transient organizational structures. The industry comprises diverse trades with distinct safety requirements and risk exposures, making it difficult to conceptualize the “aggregated” perception within the safety climate definition. Research has demonstrated significant differences in safety perceptions across trades [35]. These trade differences can be masked when aggregating safety climate data at the organizational or project levels [34].

Despite these efforts, there is still a lack of empirical evidence for fluctuations in the safety climate in the construction project context. Such evidence must account for the unique nature of various trades, immediate workgroups, subcontractors, general contractors, and project owners, who collectively define and enforce safety requirements. While the overarching definition of a safety climate highlights common themes that make the concept applicable across multiple industries, it is crucial to address the distinctive characteristics of the construction industry when conceptualizing and interpreting the safety climate in this context.

2.2. Measuring the Construction Safety Climate

Over the past decades, numerous studies have been conducted to measure the construction safety climate. However, the only consensus regarding its dimensionality is that it is measured using multiple scales [28,36]. These scales provide valuable diagnostic information to identify and address potential shortcomings in safety management [37]. Despite this utility, the number and contents of the scales vary widely across research, reflecting the conceptual ambiguity surrounding the construct [38,39]. Notably, most scales used in published research have been applied only once and are rarely cited in subsequent studies [2]. Safety climate scales can generally be categorized into two groups: general or universal scales applicable across contexts and industry-specific scales designed to capture nuanced, context-specific factors [40]. Although industry-specific scales have been shown to be more predictive of safety behavior and risk assessments than universal scales [41], very few studies have explicitly developed industry-specific scales. Consequently, universal scales remain the most commonly used in construction safety research [28,36].

Another challenge in measuring the safety climate in construction is due to its dynamic nature [42]. The construction work environment constantly evolves in topology and work conditions [43]. The concurrent activities of various trades further complicate capturing the shared perceptions of a shifting workforce [2]. These dynamics suggest that longitudinal measurement approaches are more suitable for assessing the safety climate, as they can capture its fluctuations over time [20]. By providing a comprehensive picture of these variations, longitudinal measurements enable project management to allocate safety resources effectively and proactively mitigate potential risks. Nevertheless, most existing construction safety climate studies have relied on cross-sectional designs [36].

Lastly, traditional safety climate measurement predominantly employs questionnaire-based data collection. Respondents are typically asked to evaluate safety practices using Likert scale responses, with shared perceptions aggregated as the average of individual responses. Such quantitative approaches establish a baseline to statistically validate the causal relationship between the safety climate and safety performance indicators. However, the meaning and interpretation of such a safety climate score can be nonintuitive since it does not answer the questions of “why” and “how” in safety management [21]. There has been a growing call for mixed-methods research designs that integrate quantitative and qualitative methodologies to improve the validity and reliability of the resulting data [44,45].

3. Research Methods

In response to the conceptual gap in the dynamic facets of the construction safety climate and the methodological gap in measuring it, this study employed a longitudinal mixed-methods approach to provide insights into construction site safety management. This approach offers several advantages. The longitudinal design captured temporal changes in safety climate perceptions, providing insights into how these perceptions evolve across project phases. By incorporating both qualitative and quantitative data, this study captured trade-specific and contextual factors that influence safety climate perceptions, providing a nuanced understanding of safety climate dynamics. Additionally, the application of Item Response Theory (IRT) enhanced the precision and reliability of safety climate measures by identifying survey items with high sensitivity to dynamic changes. The development process and the detailed data collection procedure are presented below.

3.1. Safety Climate Evaluation Protocol Development

To evaluate the dynamics of the construction safety climate, this study followed a structured process comprising the following steps:

• Step 1: Identify the universal scales

Based on a comprehensive literature review, seven commonly used safety climate scales were selected: management commitment, safety communication, workers’ safety involvement, safety training and education, safety rules and procedures, work pressure, and supportive environment.

• Step 2: Develop construction industry-specific scales

Building on previous work [46], a structured workshop was conducted with 17 national construction safety professionals to develop construction industry-specific scales. Through a collaborative process, this workshop aimed to identify, organize, and prioritize safety topics unique to the construction context. Activities included brainstorming potential safety topics, grouping and prioritizing these topics, and evaluating their relevance to and reliability for construction workers.

• Step 3: Validate the survey-based process

A pilot study was conducted to validate the questionnaire. Forty-eight construction workers from a facility management organization completed the survey and provided feedback on the clarity of the questions, response options, and any difficulties encountered. Five questions were excluded from the original questionnaire draft based on the feedback, yielding twenty-five questions in the final version, including eighteen universal safety climate questions and seven construction industry-specific questions, as shown in

Table 1. Safety climate survey questions.

Scale Level	Scales	Number of Question Items	Selected References
Universal	Management commitment	3	[21,47,48]
	Safety communication	3	[18,42,49]
	Workers’ safety involvement	3	[21,47,49]
	Work pressure	3	[48,49]
	Supportive environment	3	[47,48,49]
	Safety training and education	3	[18,21]
Construction industry-specific	Adequate time to plan for a task	1	Self-developed
	Awareness of upcoming tasks	1
	Awareness of job site changes over the next period	1
	Awareness of hazardous materials on-site	1
	Communication with other trades	1
	Access to safety resources	1
	Workplace congestion	1
TOTAL		25

. Question items were developed based on the scales. The complete list of the 25 safety climate questions is presented in Appendix A.

• Step 4: Validate the overall process

A focus group was conducted to refine the questionnaire and develop the measurement process. Focus groups are widely used to gather and explore diverse perspectives [50]. Fifteen safety professionals from national construction firms participated in the discussion. Three major stages of the measurement process were proposed, including surveys, follow-up interviews, and result analysis and sharing. Participants were asked to provide suggestions on this process, the value they would foresee, and concerns about implementing the method.

3.2. Data Collection Procedure

This study employed a longitudinal and mixed-methods approach to evaluate the dynamics of the safety climate in construction, as illustrated in Figure 1. Two rounds of data collection were conducted with an interval to capture fluctuations in safety perceptions as the project progressed. Each round of data collection involved three stages: (i) Pre-data collection from a safety audit observation, (ii) Mixed-methods data collection via a questionnaire survey and semi-structured interviews, and (iii) Post-data collection with data analysis and results sharing. This integrated approach provided a comprehensive perspective on safety climate dynamics across multiple project phases.

3.2.1. Case Study Context

The study selected commercial projects in their early stages to implement the proposed approach to evaluate the dynamics of the safety climate. The commercial projects in their early stages were selected for their substantial variability in site conditions, trade-specific tasks, and workforce interactions, making them ideal for assessing fluctuations in the safety climate. Additionally, they offered a unique opportunity to develop proactive safety strategies to enhance outcomes as the project progressed. The case study approach provided an in-depth understanding of safety climate dynamics within the specific context of a commercial construction project [51]. The case study project was a six-story, 95,737-square foot dormitory construction project located on a U.S. university campus. The design-build project was scheduled to take 18 months and cost approximately 35 million dollars. This study employed a longitudinal design, with data collection at two distinct time points: T1 (during the structural erection phase) and T2 (during the interior finishing phase). This approach allowed us to observe changes in safety climate perceptions as the project progressed and site conditions evolved.

3.2.2. Data Collection Process

Each data collection round followed a structured process detailed as follows:

(1)

Observation: The observation phase was conducted 1–2 days before the questionnaire was administered to document site-specific safety conditions, such as housekeeping, equipment setup, temporary facilities, and worker safety behavior. These observations provided critical contextual information to inform subsequent survey and interview data collection. At T1, five safety issues were identified, including an unsecured main gate, improper use of lock-out/tag-out lock, a worn blade on the metal stud saw, improper storage of combustible material, and unsafe ladder usage by an electrician. By T2, the site’s safety conditions were improved significantly, with no recordable safety issues observed.

(2)

Questionnaire Survey: A paper-based questionnaire survey was distributed directly to construction workers during a safety luncheon following the observation phase. This allowed high participation and immediate data collection. The questionnaire included the following four sections: (1) Demographic information on respondents’ project role, trade, and years of experience, which allowed analysis by job type and experience level; (2) Respondents’ main working areas on-site to assess spatial dynamics; (3) Twenty-five safety perception statements based on a Likert scale ranging from one to five, where one means strongly disagree and five indicates strongly agree; and (4) Open-ended questions to encourage respondents to comment on specific safety concerns and anticipated challenges, allowing qualitative insights into safety issues not covered by the quantitative measures. The sample size in T1 was 52 and increased to 131 in T2, reflecting the workforce increase as the project advanced.

(3)

Semi-structured interviews: Semi-structured interviews were conducted with the crew leaders across different trades after preliminary data analysis with the survey data. Semi-structured interviews balance standardization for comparability and depth for uncovering detailed insights into safety-related factors [52]. Participants were selected through purposive sampling to ensure representation across various trades. Criteria included active participation in the project during the data collection phases and willingness to participate in the study. The sample size was determined based on the principle of data saturation, where no new themes emerged from additional interviews [53]. This sample size ensured representation across key trades and roles, providing diverse perspectives on safety climate perceptions. The interviews lasted between 20 and 40 min depending on the depth and scope of the participants’ responses. Task-specific crew leaders required approximately 20 min, while supervisors and management personnel, due to their broader oversight, needed up to 40 min. The interviews began with questions about the participants’ trade and years of construction experience. Preliminary analysis results from the survey were used as prompts for the interview questions to capture the perspectives of the crew leaders on safety-related topics within the project context. The structure of the questions for the interview is presented in Appendix B. The semi-structured interviews provided a consistent flow of ideas to compare across multiple trades, and the participants were allowed to offer new directions of conversation based on specific topics, generally related to the specifics of their trades.

3.2.3. Preliminary Analysis and Results Sharing

After each data collection round, preliminary results were shared with the project management team. This collaborative process enabled the refinement of safety procedures and informed planning for the subsequent phases of the project. By implementing this mixed-methods approach across two distinct time points, the study effectively tracked changes in safety perceptions and site conditions over time, offering a comprehensive multi-wave analysis of the safety climate within a dynamic construction environment. A summary of the two data collection rounds is provided in

Table 2. Data collection summary.

Phase	Observation	Questionnaire-Based Survey	Semi-Structured Interview
T1	Five safety issues	52 responses were collected (43 valid responses after data screening)	4 participants
T2	None	131 responses were collected (108 valid responses after data screening)	12 participants

.

3.3. Data Analysis

The data analysis was designed to ensure survey responses’ quality and reliability, validate the instrument, and extract meaningful insights from the qualitative data.

3.3.1. Data Screening and Treatment

Before analysis, the quality of the survey responses was checked. Of the 183 completed questionnaires, 32 were excluded from the data set due to either more than 5% missing responses or systematic response patterns (e.g., identical ratings for all questions), which could compromise data integrity. For the remaining data set, the Expectation Maximization (EM) method was used to handle any remaining missing values using SPSS 27 [54], thus ensuring a complete data set for further analysis.

3.3.2. Internal Reliability

The internal reliability of the safety climate survey was assessed using Cronbach’s alpha. The overall internal reliability score was 0.901, indicating good internal consistency among the survey items, which supports the reliability of the safety climate measures [55].

3.3.3. Discriminant Validity

Discriminant validity refers to the extent to which two constructs are distinct and uncorrelated, ensuring the measurement is specific to the intended variable [56]. This study employed an Item Response Theory (IRT) approach to confirm the discriminant validity of the constructs. IRT is widely used in the fields of psychology, education, and buildings to analyze various testing results [57,58,59]. It is a robust statistical approach that helps analyze questionnaire items’ performance based on latent traits, enabling precise measurement and comparison across populations. IRT analysis is widely applied in safety climate research to streamline survey questions [60], evaluate safety climate scale information at specific item levels [61], and validate safety climate scores [62,63]. This study applied IRT to identify the most sensitive and reliable safety climate survey items for measuring perceptions. The Graded Response Model (GRM) was applied to analyze the safety climate survey items, as described by Equation (1). GRM is a common IRT application, particularly suited for ordinal data like Likert scale responses, to assess items’ discrimination power and difficulty levels. The GRM model calculates the probability of a respondent endorsing a particular category or higher based on their latent trait level (θ). The probability $P_{ik}$ is calculated as:

$$P_{ik}={\displaystyle \frac{1}{1+e^{-a_1(\theta -b_{ik})}}}$$

(1)

where:

• $a_i:Discriminant parameter for item i,$
• $\theta :Latent trait level of the respondent,$
• $b_{ik}:Difficulty parameter for item i and category k$
• $e:$ Euler’s number.

By analyzing the discrimination parameters, this study identified key survey items that effectively differentiated between varying safety climate perceptions, offering new insights. The IRT analyses were performed with the R software Version 4.3.2 [64] open-source package LTM (Latent Trait Modeling).

3.3.4. One-Way ANOVA

One-way analysis of variance (ANOVA) is a statistical method used to determine whether there are statistically significant differences between the means of three or more independent groups [65]. This study employed one-way ANOVA to evaluate differences in safety climate perceptions across trade groups and project phases. ANOVA compares between-group and within-group variability, as expressed by the F-statistic (Equation (2)):

$$F={\displaystyle \frac{Between-Group Variance}{Within-Group Variance}}$$

(2)

where F is the F-statistic, calculated by dividing the variance among group means by the variance within the groups.

This study used ANOVA to evaluate differences in safety climate perceptions across project phases (T1 and T2) and trade groups. Additional analyses were conducted to interpret significant differences across trades to address potential sample imbalance. While ANOVA is typically not used in longitudinal designs with repeated measures, it was applied to analyze between-group differences in safety climate responses across distinct phases and trades rather than tracking the same individuals over time. By combining IRT for item-level sensitivity and ANOVA for group-level differences, the study achieved a comprehensive understanding of both the measurement instrument and contextual factors influencing safety climate perceptions.

3.3.5. Qualitative Data Analysis

Thematic analysis was used to analyze qualitative data from semi-structured interviews [66]. The procedures we undertook included the following four steps: (1) Transcription: interview data were transcribed for analysis; (2) Coding development: Coding categories were established based on the research purpose and themes emerging from the interviews, such as perceived trade risks, project logistics, schedules, rules and procedures, communication, and recognized hazards; (3) Coding: The segments of text were assigned to the predefined categories; and (4) Pattern identification: The coded data were reviewed to identify patterns and relationships within the qualitative insights.

4. Results

4.1. Demographics

Demographic data were collected through the survey, including respondents’ roles, trades, and years of working experience in construction. The demographic characteristics are summarized in Figure 2, which highlights fluctuations in the workforce demographics between T1 and T2, reflecting the evolving project phases. At T2, the proportion of workers decreased from 70% to 65%, while supervisors increased from 20% to 25%, indicating heightened oversight in later stages. Respondents with less than two years of experience rose from 15% to 20%, reflecting an influx of newer workers. Trade representation also shifted, with carpentry declining from 40% to 30% and electrical and plumbing increasing, consistent with the transition from structural to finishing work. The overall proportion of carpenters in the sample was higher than those in other trades. This reflects the nature of the project during the data collection period, as carpentry activities were predominant in these phases of construction. These changes in respondents between T1 and T2 emphasize the dynamic nature of construction projects and their influence on safety climate perceptions.

4.2. Overall Safety Climate Evaluation

The survey results highlighted high- and low-scoring items, reflecting the workforce’s positive and negative safety perceptions. As shown in

Table 3. Safety climate items with the highest and lowest average scores, derived themes, and example quotes.

Survey Items	Average Scores	Themes	Exemplary Quotes from Interviews
Positive safety perceptions
The management seriously considers any workers’ suggestions for improving safety. (Item 1)	4.65 (T1)	Safety policy	“We raised it and worked it out with them [the general contractor] to amend the safety policy to permit us with the ladder use; that’s what we needed”.
There are regular safety meetings/inspections at this job site. (Item 13)	4.47 (T2)	Safety planning	“The only thing we do is awareness, which is done daily. We discuss such issues in the morning when we review our PTSA (pre-task safety analysis) as a group”.
			“Just in the morning stretches, usually we have little safety talk to identify the hazards of the day”.
		Informal safety talks	“There is a perfect thing: the morning stretches. A lot of good information gets pulled out of there”.
Negative safety perceptions
Sometimes, the workplace conditions can hinder my ability to work safely. (Item 20)	2.44 (T1)	Trade-stacking	“I think it’s a very confined workspace. I think there are a lot of trades working close to each other, so that’s a big risk there”.
			“I got a lot of crane activities out there based on what I do. Sometimes, people get in the way”.
		Material handling	“I think the biggest thing is the ability to receive deliveries… something as simple as unloading a truck is difficult because of how congested it is. We have to worry about struck-by or caught-in-between [risks]”.
			“We can’t do our jobs as it [material storage] hinders us doing our jobs when those materials are stored in the room that they tried to work in. Ultimately, that leads to the housekeeping, so it’s a safety thing”.
Some safety rules and procedures are not very practical. (Item 4)	2.74 (T2)	Safety rules and procedures	“We use screw guns to put screws in. Sometimes, very often, the screws will catch on their gloves, and it hurts. And the gloves should be off when those screwing with screw guns; or fingerless gloves…Yet fingerless gloves defeat the purpose of cut-level gloves”.
			“A couple of years ago, fall protection was not a big thing… But I agree with fall protection 100% because you will never know what could happen”.
		Willingness to accept safety rules	“I’m a big fan of stilts, but it’s prohibited on this project. I don’t want to use scaffolds, but that is the rule”.

, the highest rated item at T1 was ‘The management seriously considers any workers’ suggestions for improving safety’, reflecting strong trust in management commitment. Conversely, the lowest rated item at both T1 and T2 was ‘Sometimes, the workplace conditions can hinder my ability to work safely’, indicating ongoing concerns about workspace constraints.

4.2.1. Positive Safety Perceptions

At T1, the highest rated item was “The management seriously considers any workers’ suggestions for improving safety (Item 1)”. This reflects strong trust in management’s commitment to safety. For example, an electrician crew leader noted that the general contractor adjusted ladder safety rules based on worker input, demonstrating responsiveness. At T2, the highest rated item was “There are regular safety meetings/inspections at this job site (Item 13)”. Interviews identified the following two key themes: “safety planning” and “informal safety talks”. Proactive measures, such as daily pre-task safety analyses and informal morning stretches, were highlighted as effective tools for addressing emerging safety concerns.

4.2.2. Negative Safety Perceptions

The lowest scoring item at both T1 and T2 was “Sometimes, the workplace conditions can hinder my ability to work safely (Item 20)”. This item’s low score, flipped due to its negative wording, reveals widespread concerns about unsafe conditions. Key themes included “manpower congestion”, which intensified during interior work at T2, and “material storage congestion”, which was exacerbated by uncoordinated deliveries. Another low-scoring item was “Some safety rules and procedures are not very practical”. The themes were “safety policy conflicts with production” and “willingness to accept safety rules”. Themes included conflicts between safety policies and production needs, as well as resistance to safety rules. For example, some workers found mandatory cut-level gloves burdensome for specific tasks, such as carpenters using screw guns. Senior workers also expressed reluctance to adopt new safety protocols, highlighting the need for clearer communication and targeted training.

4.3. Dynamics of the Safety Climate

4.3.1. Change Trend of the Safety Climate over Time

As the project transitioned from structural erection (T1) to interior-focused activities (T2), the physical environment and workforce density changed significantly. However, one-way ANOVA revealed no statistically significant differences in the overall safety climate between T1 and T2, suggesting stability across project phases. Nevertheless, qualitative insights revealed a shift in specific safety concerns in two aspects. For instance, outdoor tasks at T1 were impacted by snow and ice, while T2 saw increased risks due to workforce congestion during interior tasks. Although these conditions did not result in major safety incidents, interviewees noted that congestion elevated minor risks, emphasizing the need for proactive logistics management.

4.3.2. Differences by Trade

One-way ANOVA was performed to assess whether safety climate perceptions were statistically different among the construction trades. As presented in

Table 4. Statistically significant differences in safety climate perceptions across all trades.

Question Item	F Value	p Value
I have enough space to perform my work. (Item 7)	2.12	0.02
I feel comfortable correcting other trades when they are not following the safety rules. (Item 25)	2.12	0.02
Written work procedures match how tasks are done in practice. (Item 8)	2.77	0.03

, while most items stayed stable across the twelve trades, three were identified as statistically different, with p-values smaller than 0.05. The first item was “I have enough space to perform my work (Item 7)”. The interviews validated this. The sheet metal workers and the site workers claimed that the space congestion hindered their work production. Deliveries and storage of materials were the most common difficulties that they encountered daily. Equipment operators were also experiencing problems due to the limited room they had to move around. For other trades, like painters and carpenters, the workspace was sufficient for them to perform their tasks. The second item was “I feel comfortable correcting other trades when they are not following the safety rules”. The cement trade was more willing to correct coworkers’ unsafe behavior in other trades. In contrast, the demolition trade hesitated to speak up when coworkers not from their crew were not following the safety rules. The third item was “Written work procedures match how tasks are done in practice”. The electricians agreed with this statement more than the excavation workers. As the crew leader stated during the interview, the excavation workers were experiencing more unforeseen site conditions and weather challenges This might lead to on-site adaptations of the planned work procedures.

During the interviews, we asked the crew leaders about their perceived risk levels of their tasks. The perceptions varied; for example, the asbestos abatement trade commented, “All trades have high risk, but people in our industry would say…super high because they don’t understand how we do it”. In contrast, although electricians’ risk level was considered high according to historical data on safety accidents, electrician crew leaders perceived a moderate risk level. He stated, “With the advancement of safety that has been made industry-wide, our level of risk has been reduced, mainly through training and technology improvements for electrical work. With those factors, I would say our risk level is no greater than any other trade”.

4.4. Insights from the Item Response Theory (IRT) Analysis

IRT analysis identified three items with high discrimination parameters, indicating their sensitivity to changes in safety climate perceptions. These items were used as prompts in T2 interviews to explore nuanced workforce perspectives.

Table 5. Safety climate items with the highest discrimination parameters ¹, the derived themes, and example quotes.

Survey Items	Discriminant Parameter	Themes	Example Quotes
I feel that my supervisors and other top managers care about my safety. (Item 18)	3.29	Safety team	“We are fortunate to have a good safety team and representation on the job, and safety here is proactive”. “They [the construction management] haven’t been pushing production over safety”.
		Priority of safety	“There are no contractors like to be late, but if it’s between being late and doing something unsafe to hurt my guys, we are going to be late”.
		Safety training	“The biggest improvement is keeping the guys trained in safety”.
I am informed that changes to my job site over the next period will impact me. (Item 19)	2.74	Typology changes on-site	“There are always changes, and that’s big with us [site workers] because a lot of our work is underground…you don’t know things are there until you find them”.
		Weather conditions	“Weather plays a big factor in what we do”.
I feel comfortable correcting other trades when they do not follow the safety rules. (Item 25)	2.05	Communication with other trades	“We have a very good job here regarding safety, in my opinion”.
			“I guess if there were risks, I would say something to them or at least let them know they should do something different. And if they choose not to, that’s on them”.
		Coordination among trades	“Just coordination helps to make it safe”.

¹ The discrimination parameter was a key output of the IRT analysis. Higher values of the discrimination parameter indicate that an item was more sensitive to differences in safety climate perceptions among respondents.

summarizes these items, along with their derived themes and illustrative quotes from the interviews. The safety climate item with the highest discrimination parameter was “I feel that my supervisors and other top managers care about my safety (Item 18)”. Three broad themes were identified from this item during the interviews. The first theme, “safety team”, was acknowledged as there were dedicated safety representatives on the project, and they took proactive safety management measures. The second theme was “priority of safety”. From the subcontractors’ perspective, the general contractors prioritized safety over production. From the crew leaders’ perspective, they claimed to allow the crew sufficient time to work safely. The third theme was “safety training”. Several interviewees commented on the importance of training to improve safety, even for senior workers, who need to understand the updated safety requirements despite their extensive working experience on construction sites.

The second item with a high discrimination parameter was “I am informed that changes to my job site over the next period will impact me. (Item 19)”. Changes in site typology and weather impacts emerged as major themes. Site workers raised concerns about potential hazards due to unforeseen underground conditions, while others, like roofers and cement workers, noted that fluctuating weather conditions impacted their tasks, underscoring the importance of adaptive safety communication.

The third item with a high discrimination parameter was “I feel comfortable correcting other trades when they are not following the safety rules (Item 25)”. While most workers were comfortable correcting unsafe behavior, one interviewee mentioned that peer enforcement might be met with resistance, underscoring the complexities in inter-trade communication. This feedback underscores the need for fostering a collaborative safety culture, especially in close-quarter work settings. The other theme was “coordination among trades”. The case study project had a very confined space for construction, especially during T2, so coordination of schedules was recognized as an essential measure to improve safety.

5. Discussion

5.1. Key Findings

This study aimed to evaluate the safety climate within the dynamic context of construction projects. The results revealed both stable and fluctuating safety climate features as perceived by workers across multiple trades and project phases.

The survey results indicated a consistent overall safety climate throughout the research despite evolving site conditions and building activities. This finding contradicts a prior study exploring the dynamics of a safety climate based on survey results [20]. The stability of the overall safety climate can be attributed to the general contractors’ proactive safety policies and practices, ensuring that the workforce perceived a stable and positive safety climate. The findings suggest that consistent organizational safety policies can mitigate the impact of work environment dynamics.

Despite maintaining a stable safety climate, qualitative data revealed distinct safety concerns that varied according to the trade and project phases. During transition phases, such as moving from structural to interior work, workers reported heightened safety concerns due to shifting site conditions, overlapping trade activities, and new operational risks. These findings highlight the dynamic nature of the safety climate and the importance of adapting safety strategies to project phase-specific challenges. Safety climate perceptions also differed across trades and were influenced by trade-specific risks. For instance, carpenters expressed more concerns about physical risks related to heavy machinery and site congestion, while electricians emphasized the need for improved coordination with other trades. Contextual factors such as workforce composition, site logistics, and management practices also played critical roles in shaping safety climate perceptions, underscoring the importance of considering both trade-specific and contextual elements in safety management.

The integration of IRT and ANOVA allowed a multi-level analysis of safety climate dynamics. IRT identified the survey items most sensitive to safety climate variations, while ANOVA highlighted significant group-level differences, offering actionable insights for trade-specific and phase-specific safety strategies.

This study highlights the value of a longitudinal mixed-methods approach in providing a nuanced understanding of safety climate dynamics. This approach captured temporal changes and integrates both quantitative and qualitative data, allowing a deeper exploration of trade-specific and contextual influences. The results demonstrate that such an approach can inform targeted interventions and more effective safety management strategies.

5.2. Comparison with Existing Practices

The findings align with existing literature emphasizing the importance of the safety climate in construction projects while advancing the understanding of safety climates by addressing their dynamic nature. The longitudinal mixed-methods approach used in this study provided a more nuanced understanding of safety climate dynamics. Emerging methodologies such as data mining have demonstrated significant potential for understanding construction safety and hazard awareness [67]. While these approaches capture large-scale trends, our longitudinal mixed-methods design complements them by focusing on trade-specific and contextual safety climate perceptions within construction projects, providing actionable insights for tailored safety management strategies.

Additionally, while previous studies have identified management commitment as a critical factor for a safety climate, this study emphasizes its role in trade-specific contexts. For example, safety leadership and communication impacted trades working in high-risk environments or during transition phases. These findings offer practical insights that extend beyond generalized safety protocols.

5.3. Practical Implications

The results of this study highlight the importance of the safety climate not only as a key indicator of pre-safety management but also as a dynamic tool for addressing trade-specific and phase-specific safety challenges. The stability of safety climate perceptions observed in this study underscores the roles of consistent management commitment and proactive safety policies in mitigating the effects of dynamic site conditions. Additionally, trade-specific concerns, such as physical risks for carpenters and inter-trade communication for electricians, highlight the need for tailored safety strategies. Rather than implementing universal safety management strategies across all trades, project management could establish targeted safety measures to address specific safety concerns associated with each trade during various phases. For example, safety training programs should be customized to address the unique risks faced by each trade. Carpenters may require enhanced physical risk mitigation strategies, while electricians might benefit from improved inter-trade communication protocols.

The longitudinal findings highlight the importance of continuous safety climate assessments throughout project phases. Periodic safety surveys, combined with real-time monitoring of site conditions, can help managers identify and address emerging risks as a project evolves. The findings also suggest that safety climate assessments should be used as a proactive monitoring tool to identify and address emerging risks during transitional project phases. By leveraging advanced technologies such as Building Information Modeling (BIM), wearable sensing devices (WSDs), and augmented reality (AR), construction project managers can enhance the precision of safety monitoring and implement adaptive safety interventions in real time [68,69]. Furthermore, the critical role of management commitment in maintaining a stable safety climate underscores the need for consistent safety leadership and communication throughout a project.

6. Limitations and Future Work

This study has several limitations. First, it focused on new commercial building construction projects in the early phases because these stages are highly dynamic, involving significant workforce interactions, evolving site conditions, and high-risk activities. However, the study was constrained by the availability of a small number of projects, which resulted in a small sample size. This constrains the broader applicability of the results. Second, due to a lack of pre-defined metrics of site dynamics, it is challenging to statistically link site dynamics to the safety climate. While the information guided project management in developing safety plans and risk management, no statistical model is available to link the data to perceived safety. Lastly, this research focused on individual perceptions of safety management within a specific context. Other mediating factors, such as educational background, knowledge, and experience, may impact the safety climate but are challenging to measure and validate dynamically. Demographic differences between the two data collection samples, including average years of work experience, may have impacted the results. Controlling for individual knowledge and experience as variables is particularly challenging in the construction industry, where trades and workforce compositions constantly change throughout a project.

Future research could address the limitations of this study by expanding the scope to include a broader range of construction projects, such as residential and infrastructure, across different geographic regions. While this study focused on contextual and trade-specific factors, incorporating individual-level factors such as education, cultural background, and risk tolerance in future research could provide a more comprehensive understanding of safety climate dynamics. Future studies could leverage digital tools such as BIM or wearable sensors to monitor safety climates in real-time, providing more detailed and dynamic information to complement traditional survey and interview methods.

7. Conclusions

Previous studies have underscored the importance of a safe climate as a leading indicator in proactive safety management, particularly in high-risk industries like construction. The complexity and dynamic nature of construction projects make it challenging to conceptualize and measure safety climates in construction. This study employed a longitudinal, mixed-methods approach to evaluate the safety climate within a residential building construction project. The findings offer empirical evidence for the complex dimensions of the safety climate in the dynamic work environment of construction projects.

The longitudinal design was used to explore changes in safety perceptions across different phases of the project. The results demonstrated that the project maintained a consistently strong safety climate despite evolving site conditions and varying construction activity. This implies the critical role of a general contractor’s proactive safety strategies in fostering a stable sense of safety among construction workers, even though they come from different subcontractors and are involved in tasks with varying risk levels. However, qualitative analysis indicated evolving safety concerns at different stages of the project. The variations in safety climate perceptions among the multiple trades indicated the need to develop safety management practices tailored to each trade’s unique context. For example, when a safety policy affects a particular trade, it should be explicitly communicated during those workers’ orientation.

This study highlights the role of the safety climate as a dynamic tool for designing and implementing targeted safety interventions for specific trades and project phases. These findings offer actionable insights for improving safety outcomes in high-risk construction environments and highlight the importance of proactive safety management and integrating advanced technologies to improve safety outcomes in high-risk construction environments. The longitudinal, mixed-methods approach also contributes to the theoretical understanding of a safety climate by framing it as a dynamic construct.

The findings for the stability of safety perceptions highlight the importance of a general contractor’s safety policies. At the same time, the dynamics of the safety climate imply the need to address trade-specific safety concerns and develop flexible safety strategies. This could involve ergonomic improvements like planning site layouts and improving material storage, which can help mitigate risks in crowded work environments and enhance overall site safety.