The mediating role of risk perception in the relationship between chemical safety knowledge, GHS awareness and safety behavior Authors: Sayed Vahid Esmaeili, Reza Esmaeili, Ali Mohammadi, Marzieh Baharlouei, Mahdi Jalali, Amir Kavousi, Mostafa Pouyakian Published: 2025 ABSTRACT AbstractExposure to hazardous chemicals remains a critical challenge in industrial settings, leading to significant adverse health outcomes and substantial economic losses annually. This study aimed to investigate the mediating role of risk perception in the relationship between chemical safety knowledge, GHS awareness, and safety behavior. This analytical modeling study was conducted among 360 workers from four chemical-enterprise scales (micro, small, medium, and large) in 2024. Data were collected using standardized questionnaires. Data analysis were performed using SPSS v.26. The mediating role was examined using structural equation modeling (SEM) in AMOS v.24, and model fit was assessed using indices such as CFI, RMSEA, and χ²/df. The findings exhibited a general lack of familiarity with less commonly used pictograms, such as toxic materials (80.56% unfamiliarity), and a high reliance on informal training in small enterprises (84.62%). Data analysis illustrated that enterprise size significantly influenced safety knowledge and awareness levels (P < 0.001). Large enterprises demonstrated the highest knowledge scores (27.08 ± 3.05) and safety behavior (3.65 ± 0.53), while micro and small enterprises scored the lowest. Furthermore, risk perception was significantly higher in large enterprises (3.94 ± 0.26) compared to small (3.65 ± 0.33) and micro-enterprises (3.33 ± 0.36; P < 0.001). Path analysis confirmed that risk perception acts as a significant mediating variable, strengthening the relationship between safety knowledge, GHS awareness, and safety behaviors (β = 0.203, P < 0.001). This study addresses a gap in the literature concerning the mechanisms through which key chemical safety variables interact and underscores the urgent need for targeted interventions, such as realistic hazard simulations, standardized bilingual labels, and digital technology integration, to enhance chemical safety across organizational scales.Keywords: Chemical safety knowledge, GHS awareness, Risk perception, Safety behavior, Structural equation modelingSubject terms: Health occupations, Statistics, Chemical safety Introduction Occupational exposure to hazardous chemicals represents a critical and pervasive global public health challenge1. The widespread use of chemicals and solvents across industrial and non-industrial sectors, while fundamental to modern production, poses severe risks to human health2. These substances are responsible for a spectrum of acute and chronic health conditions, including carcinogenicity, respiratory disorders (e.g., asthma, vaginitis), dermatological conditions (e.g., dermatitis), ocular injuries (e.g., chemical burns), and damage to vital organs such as the kidneys, liver, and cardiovascular system3,4. Statistics indicate that 18–30% of occupational diseases in Europe and 2 million global fatalities in 2021 were linked to chemical exposure5,6. Beyond long-term health effects, improper handling can lead to immediate safety crises, including fires and explosions, further threatening workers and infrastructure7,8. These realities frame chemical safety not merely as a regulatory issue, but as a shared responsibility among governments, employers, and workers9. A substantial body of research has been dedicated to understanding and improving chemical safety. Previous studies have often focused on engineering controls, regulatory frameworks, and the technical design of warning systems. In this context, international organizations such as the occupational safety and health administration (OSHA), the world health organization (WHO), and the international labour organization (ILO) have established key frameworks to mitigate chemical risks. These include the hazard communication standard (HCS), the inter-organization programme for the sound management of chemicals (IOMC), the strategic approach to international chemicals management (SAICM), and the globally harmonized system of classification and labelling of chemicals (GHS)10,11. Central to these frameworks is the mandate for employer-led worker training, designed to equip employees with the knowledge to identify hazards and manage chemicals safely. The GHS, in particular, serves as a pivotal tool by standardizing hazard classification, pictograms, and Safety Data Sheets (SDS), thereby providing a universal language for communicating risk2,12,13. These standards have directly contributed to reducing chemical-related incidents and occupational diseases14. However, the success of these frameworks hinges on training quality. Standards such as ANSI Z535 and ISO 3864, which define minimum comprehension thresholds (85% and 67%, respectively) for warning signs, demonstrate that effective signage design must be coupled with continuous training15. Moreover, individual factors including motivation, personality traits16–18, education level19, cultural differences and sign design features20 significantly influence risk perception quality, underscoring the need for training programs tailored to the diverse needs of workers. Meanwhile, exposure, familiarity, and perception of individuals toward different chemicals may vary depending on industry size, exposure nature, and occupational context21,22. Based on classifications by organizations and related studies, industries are categorized into four groups by employee count: micro-enterprises (< 10 employees), small (10–49 employees), medium (50–249 employees), and large enterprises (≥ 250 employees)23. However, it is important to note that such classifications can vary across different countries and regulatory contexts. While large industries and corporations comply with chemical safety standards through organizational oversight (e.g., Occupational Health and Safety (OHS) systems) and regular training24,25, a significant portion of the global workforce is employed in small and medium enterprises (SMEs) and self-employed roles (e.g., mechanics, painters, hairdressers, service workers), which often lack access to formal training, protective equipment, and adequate supervision26,27. In countries such as the United States, Canada, Australia, and Iran, Occupational Safety and Health (OSH) regulations for these groups are less stringent, and occupational injury reporting systems are underperforming (30). This regulatory gap leads many SME workers to acquire information from informal sources like social media, which are often inaccurate or insufficient28. Despite the presence of standardized labels and symbols on chemical substances, the utility of this information hinges on users’ accurate comprehension29,30. Studies indicate that failure to read labels31, prioritizing non-safety information (e.g., usage instructions or marketing details), or disregarding warnings even among individuals with prior knowledge32 lead to misperception of risks and unsafe and high-risk behaviors. The mechanisms by which these factors contribute to unsafe chemical exposure and health effects can be conceptualized as a chain of influence. Inadequate chemical safety knowledge and low GHS awareness directly impair an individual’s ability to accurately identify hazards. This deficiency leads to a misperception of risk (underestimating the severity or likelihood of harm) and low self-efficacy (doubting one’s capability to handle chemicals safely). Consequently, this results in poor safety behavior, such as not using personal protective equipment or mishandling chemicals, which directly increases the probability of exposure and adverse health events. In this context, two psychological factors including perceived risk (sense of vulnerability to hazards) and self-efficacy (belief in one’s ability to prevent harm) play main roles in safety behavior32,33. Behavioral models such as the protection motivation theory (PMT) and the Health Belief Model (HBM) emphasize that risk perception predicts behavioral changes34–38. However, empirical evidence in this domain remains inconsistent. Some studies report a significant relationship between risk perception and safe behaviors36,39, while others fail to confirm this association40,41. These contradictions suggest that additional factors such as individual traits, cultural context, and access to educational resources, also influence safety-related decision-making. While existing studies have examined chemical safety broadly, there is a notable gap in the literature concerning the specific interplay between chemical safety knowledge, GHS awareness, and risk perception, particularly within the context of SMEs. This gap limits our understanding of the precise psychological and knowledge-based mechanisms that drive safety behavior in high-risk, under-regulated environments. Given the critical role of chemicals in industries and the vulnerability of specific occupational groups, this study examines the impact of chemical safety knowledge, familiarity with GHS labeling, and risk perception levels on safety behavior. The central research question is: Can enhancing GHS awareness and accurate chemical risk understanding reduce unsafe behaviors in workplaces, particularly in SMEs? Moreover, this research aims to explore the interplay between risk perception and chemical safety knowledge in shaping safe practices. Consequently, the following hypotheses were investigated in this study: Hypotheses H1: safety awareness is positively associated with safety behavior Safety awareness leads to safety behavior, such as identifying and understanding potential dangers in different environments, and more significantly, in the working environment. The link between safety behavior and safety awareness contributes significantly towards developing a safety culture, where safety awareness leads people to take initiative and report potential dangers and obey safety regulations, thereby reducing accidents and injuries42,43. The critical link between safety behavior and safety awareness has been emphasized in risky industries, where ignorance of safety can lead to catastrophic consequences. Thus, more and more companies emphasize developing safety awareness with systematic training, corporate reinforcement, and technical input. Theoretical safety concepts of safety awareness converge from various safety models, such as human factors and proactive safety, into a single approach44,45. The traditional safety approach addresses mainly obedience and penalties for safety ignorance, while safety-oriented approaches emphasize recognizing employees’ value in safety efforts. This transition represents a growing understanding that rewarding behavior and fostering a supportive environment strengthen commitment and encourage safe behavior, ultimately shaping a more stable corporate culture46,47. Individual, corporate, and environmental factors drive safety awareness. Individual factors, such as age and experience, affect people’s perception of and response to dangers. The corporate environment through training, explicit communication, and commitment from top management reinforces safety48. The physical work environment also plays a crucial role in safety awareness, as a well-planned and organized workspace helps minimize hazards. While the significance of safety awareness is widely recognized, challenges like safety ignorance and resistance to safety practices persist49,50. Continuous improvement techniques, such as frequent checks and rewarding safety behavior, are essential for maintaining safety awareness. Ultimately, fostering a safety-minded culture not only protects employees but also enhances overall performance and resilience against potential dangers51,52. Though there is a lot of evidence to support this positive association, a closer look shows that awareness might not be enough in dynamic chemical workplaces where hazards change quickly as a result of things like process changes or interactions between volatile substances. Using the Protection Motivation Theory (PMT), which holds that in order to influence behavior, awareness must set off cognitive processes like threat assessment and coping effectiveness53, Recent research exposes contextual limitations, challenging oversimplified associations. For example, reactive approaches common in SMEs, where informal training results in inconsistent behaviors and increased accident risks, frequently undermine proactive awareness (e.g., anticipatory hazard identification through formal training) in high-risk settings like the chemical industry54. Because informal learning in resource-constrained environments may foster superficial awareness without translating into sustained safe practices55, it is imperative that the hypothesis be challenged empirically while taking into account moderating variables such as enterprise size and cultural norms. In order to address this, our study looks at these dynamics in chemical enterprises and offers a nuanced test of H1. H2: safety knowledge is positively associated with safety behavior Safety Knowledge Affects Safety Behavior refers to the relationship between what people know and comprehend regarding safety procedures and how they implement them in real-world scenarios, particularly in workplace environments56. This concept is crucial for fostering a strong safety culture and preventing accidents and injuries at work. Interestingly, factors such as cognitive biases, organizational culture, training practices, and leadership influence play a significant role in shaping how individuals acquire and apply safety knowledge, making this topic a key focus in occupational safety and health management57. The importance of this subject is further emphasized by its impact on accident prevention and the promotion of safer workplaces across various industries. Research indicates that when employees possess strong safety knowledge, they are more likely to follow safe practices, ultimately reducing workplace accidents and injuries. However, misconceptions about safety practices, resistance to change, and cognitive biases can create obstacles, making the relationship between knowledge and behavior more complex58. Understanding these dynamics is essential for developing effective safety management strategies that encourage lasting behavioral changes among employees. Controversies surrounding safety knowledge affects safety behavior often stem from differing opinions on behavior-based safety (BBS) approaches and the influence of cognitive biases. Critics argue that behavior-modification techniques may fail to address deeper, intrinsic motivations, leading to surface-level compliance rather than genuine safety engagement59,60. Moreover, cognitive biases such as overconfidence and normalcy bias can impair an individual’s ability to assess risks accurately, making it more challenging to establish a proactive safety culture61. Overcoming these challenges requires a deep understanding of both cognitive and organizational factors that influence safety behavior. Though there is a lot of evidence to support this positive association, a more thorough analysis reveals contextual issues and empirical contradictions that call into question its universality, especially in high-risk chemical industries. For example, workers in SMEs frequently underestimate the risks associated with routine chemical handling, resulting in inconsistent safety behaviors. This illustrates how cognitive biases, such as overconfidence in familiar hazards, can perpetuate knowledge gaps despite prolonged exposure62. Research specific to chemicals demonstrates that safety knowledge only improves performance when mediated by elements like attitudes, leadership, and workplace climate. Additionally, knowledge transfer often fails in resource-constrained settings where informal or inadequate training predominates, undermining the direct link to behavior29,63. To provide a nuanced empirical test of H2, this challenges the hypothesis by highlighting the necessity of integrated, context-aware interventions that address organizational barriers, such as enterprise size and resource constraints, as investigated in our study. H3: risk perception mediates the relationship between safety knowledge, safety awareness, and safety behavior Risk perception mediates the relationship between safety knowledge, safety awareness, and safety behavior, playing a crucial role in how individuals interpret and respond to potential hazards in various environments64. This complex interplay highlights the significance of understanding safety protocols, recognizing risks, and cultivating appropriate behaviors to enhance workplace safety65. As organizations increasingly prioritize employee well-being, comprehending the dynamics between these elements becomes essential for fostering a culture of safety across diverse industries, including construction, healthcare and manufacturing66. Safety knowledge encompasses the information and understanding individuals have about potential hazards and risk mitigation strategies, while safety awareness refers to their ability to recognize and respond to these risks67. Research indicates that higher levels of safety knowledge and awareness can significantly enhance an individual’s risk perception, leading to improved safety behaviors and adherence to safety protocols3,68. Conversely, inadequate safety knowledge and awareness can result in misperceptions of risk, ultimately leading to unsafe practices and increased incident rates. Moreover, there are varying factors influencing perception of risk, such as cognitive and emotional factors, social factors, and organizational culture69,70. The emotional status of individuals can also heavily influence their perception of risk, with positive ones being able to induce proactiveness towards safety, and negative ones inducing avoidance behavior. Social factors, such as peer perceptions and leaders’ commitment, also play a major contribution towards affecting perceptions of risk, and therefore safety behavior in organizations. Understanding the process of risk perception mediates in developing effective safety interventions71. Organizations focusing on safety knowledge improvement, developing safety awareness, and promoting open communication are more capable of reducing risk and improving safety performance, therefore securing employees’ well-being and operational integrity52. Although there is a lot of evidence to support this mediation, a more thorough analysis using psychometric paradigms raises doubts about its simple application in chemical contexts due to difficulties and contradictory results33. Risk perception does not always accurately mediate behavior, for example, as evidenced by health behavior analogies like vaccination studies. This is especially true when familiarity biases or dread factors cause chronic chemical hazards like carcinogenicity to be underestimated36. By addressing moderating factors such as enterprise size, which our data supports (P < 0.001), we challenge the hypothesis. These factors may strengthen mediation in large enterprises with strong training, but they may weaken it in SMEs with limited resources where subjective risk assessments predominate. We also suggest future longitudinal studies to monitor the evolution of risk perception over time, offering a nuanced empirical test of H3 in dynamic chemical safety environments, in order to address the causality issues inherent in cross-sectional designs. Conceptual research model Participations in distinct groups with elevated perceived risk levels may exhibit heightened motivation to avoid risks72. However, these individuals demonstrate divergent knowledge, attitudes, and behavioral patterns contingent on varying degrees of self-efficacy beliefs. The present study seeks to provide insights into risk perception and safety behaviors among employees in large and small enterprises exposed to chemicals, consumers of chemical products, and factors influencing occupational incidents and health hazards, such as chemical safety knowledge and familiarity with GHS labeling. To achieve this, the study investigates a set of factors that have been underexplored in prior literature. Consequently, by evaluating the aforementioned hypotheses, the study examines the proposed model (Fig. 1), which integrates risk perception, behavioral drivers, and contextual factors to elucidate safety outcomes in chemical-related settings. Hypothetical and conceptual research model. Materials & methods Study design This study employed a cross-sectional, analytical modeling design to investigate the relationships between chemical safety knowledge, GHS awareness, risk perception, and safety behavior. The research was conducted over a seven-month period (July 1 to January 30, 2024) across a diverse spectrum of occupational settings in the Isfahan, Tehran and Ghazvin province. The study population encompassed workers from chemical industry such as a detergent manufacturing company, as well as workers from small workshops and urban retail shops with documented exposure to hazardous chemicals (e.g., cleaning supply retailers, laboratories, greenhouses, and automotive repair garages), which were categorized into four distinct enterprise scales (micro, small, medium, and large). To ensure high-quality data collection and participant comprehension, a team of three certified experts in chemical safety and occupational health administered the questionnaires. The data collection was performed exclusively through in-person, face-to-face sessions. This method was chosen to maximize the response rate, allow for immediate clarification of any participant questions, and ensure the integrity of the data collection process. Data collection commenced only after securing informed written consent from each participant and obtaining formal permission from the management of the involved enterprises and workshops. Inclusion and exclusion criteria The criteria for participant selection were strictly defined to ensure a homogeneous and relevant sample while minimizing confounding factors. Inclusion criteria comprised: (1) A minimum of one year of continuous employment in a role with direct or routine indirect exposure to hazardous chemicals, (2) no secondary occupations outside the chemical-related field to prevent the confounding influence of different safety cultures and practices from other industries and (3) no uncorrected visual impairments such as myopia, hyperopia, and astigmatism that would hinder the ability to read and comprehend the questionnaire. Furthermore, participants were excluded from the final analysis based on the following: (1) any participant who chose to withdraw their consent at any stage of the data collection process, (2) questionnaires with more than 10% of data missing across the core constructs were deemed invalid and excluded and (3) providing the same answer to all reverse-scored items or contradictory answers to similar questions. Sampling method & sample size calculation In this study, the required sample size was estimated using the a-priori sample size calculator for SEM73. This tool calculates the minimum sample size needed to detect a specified effect size in SEM-based studies, considering the number of observed and latent variables, anticipated effect size, confidence level, and statistical power74,75. The calculator returns both the minimum sample size required to detect the effect and the minimum recommended based on model complexity. In this research, based on a 95% confidence level, 80% statistical power, 4 latent variables, 131 observed variables, and an effect size of 0.277, the calculator yielded a minimum sample size of 342. Considering a 5% dropout rate, the final target sample size was set to 360. Data collection tools Data were gathered using a comprehensive demographic checklist and four validated questionnaires. The demographic checklist captured information on age, gender, education level, marital status, work history (including job experience and exposure history), primary sources of chemical safety information, frequency of chemical product use, and primary exposure locations (e.g., production line, laboratory, storage). The primary research instruments included: the chemical safety knowledge and awareness questionnaire (MCQ), the GHS chemical labeling status assessment questionnaire, the perceptions about hazardous substances (PATHS) questionnaire and the safety behavior questionnaire (SBQ). Chemical safety knowledge and awareness questionnaire (MCQ) The MCQ developed by Fayazi et al. (2019), is a 37-item tool designed to assess respondents’ general knowledge and awareness of chemical safety2. This tool comprises 24 multiple-choice questions to assess the respondent’s general knowledge of chemical safety, and the second one 13 visual-based items to evaluate the current status of the labeling of chemicals in their workplace. Each item is scored dichotomously: a correct answer receives 1 point, while an incorrect response receives 0 points. Results are reported as percentages and frequencies of selected responses. The raw total score (range: 0–37) was converted to a percentage scale for ease of interpretation, with a maximum score of 100. Based on this percentage, knowledge levels were categorized as follows: weak ≤ 33 (≤ 12 correct answers), moderate = 33.1–66 (13–24 correct answers), and good = 66.1–100 (25–37 correct answers) chemical safety knowledge. According to Fayazi et al.’s validation study involving technicians, supervisors, and managers in chemical plants, the tool demonstrated high validity and reliability with a content validity ratios (CVR), content validity index (CVI) and Cronbach’s alpha coefficient > 0.852,78. GHS chemical labeling status assessment questionnaire The GHS chemical labeling status assessment questionnaire, developed by Fayazi et al. (2019), evaluates the current status of chemical labeling practices and individuals’ familiarity with them2. This 40-item tool, which aligned with the globally harmonized system (GHS), chemical hazard information and packaging (CHIP), and workplace hazardous materials information system (WHMIS) frameworks, comprises two sections. Section 1 consists of 25 items (12 binary items (correct/incorrect) and 13 multiple-choice items), and Sect. Hypotheses assesses recognition of 15 pictograms, symbols, and hazard-related concepts. the pictogram recognition section requires participants to identify the correct hazard for each symbol. Participants respond based on their prior knowledge of the answer options. Results are reported as percentages and frequencies of selected responses. The tool’s validity, confirmed through evaluations across diverse occupational groups, demonstrated a CVI of 0.812,78. Perceptions about hazardous substances (PATHS) questionnaire The PATHS questionnaire developed and validated by Rubin et al. (2013) assesses public perceptions the nature of hazardous substances78. According to the tool designers, individuals’ chemical risk perception levels directly influence their likelihood of exposure to hazardous chemicals. PATHS comprises 8 scales and 31 items, rated on a 5-point Likert scale: completely disagree (1), disagree (2), neutral (3), agree (4), and completely agree (5). Each scale score is calculated by dividing the sum of its item scores by the number of items within that scale. The final overall score is derived by averaging the scores of all eight scale., resulting in a total score range of 1 to 5, where a higher score indicates a higher level of risk perception. The original validation study reported satisfactory internal consistency for all scales78. Safety behavior questionnaire (SBQ) The SBQ is a 23-item instrument developed by Mahdi Nia et al. (2016) to assess safety-related behaviors79. This tool comprises 12 items measuring safety compliance (the extent to which employees adhere to established safety rules and procedures) and 11 items evaluating safety participation (voluntary behaviors that contribute to a safer work environment, such as helping coworkers or promoting safety initiatives). Responses are recorded on a 5-point Likert scale ranging from 1 (Never) to 5 (Always). As per the developer’s guidelines, Scores are calculated as ratios rather than raw totals. For each domain (safety compliance, safety participation, and overall safety behavior), the sum of item scores is divided by the number of items in that domain. Consequently, the minimum and maximum possible scores for each variable range from 1 to 5. The final overall safety behavior score is derived by averaging all 23 items. The validity and reliability of this questionnaire have been assessed and confirmed (CVR > 0.9, ICC = 0.752 and Cronbach’s alpha coefficient = 0.902), indicating strong psychometric properties for measuring safety behavior79. Data analyses In this study, descriptive statistics were used to summarize demographic data. The hypotheses of the study were investigated using the SEM approach. We employed the SEM approach rather than simple path analysis, because SEM not only examines relationships among observed variables as in path analysis, but also estimates latent constructs and their indicators simultaneously. Given that our variables such as chemical safety knowledge, GHS awareness, and risk perception were modeled as latent constructs, SEM was more appropriate. The goodness of fit of the hypothetical model was assessed using both absolute and comparative fit indices. A fit and acceptable model should satisfy both of these index’s types. The ratio of Chi-square to Degree of Freedom (χ2/df) and the root mean square error of approximation (RMSEA) were two absolute indices used to evaluate the goodness of fit of the model. Comparative indices included the normed fit index (NFI), comparative fit index (CFI), and Tucker-Lewis Index (TLI). A χ2/df value lower than 3, RMSEA (90% CI) lower than 0.08, and comparative indices higher than 0.9 were regarded as acceptable level80,81. Skew and kurtosis scores were used to investigate the distribution of all variables included in the SEM model. All data analysis was performed using SPSS software v.26 and AMOS software v.24 at a significance level of less than 0.05. Results Descriptive characteristics of participant A total of 360 participants were enrolled in this study. The cohort had a mean age of 35.99 ± 7.19 (Range = 19.00–62.00) years, with an average work experience of 10.02 ± 6.73 (Range = 1.00–30.00) years. The mean weekly working hours were 61.31 ± 13.06 h per week (range: 40.00–84.00). Regarding demographic characteristics, the majority of participants were male (87.2%) and married (78.3%). Most participants (62.2%) held an educational level beyond a high school diploma. In terms of occupational distribution, more than half of the participants (54.2%) were employed in large-sized companies, and chemical industry workers constituted the largest job category (57.8%). A significant majority (61.1%) reported daily exposure to chemicals. The detailed demographic and work-related characteristics of the participants are presented in Table 1. Table 1Demographic characteristics and work-related information of participants.VariableFrequency (N)Percent (%)GenderMale31487.2Female4612.8Marital StatusSingle7220.0Married28278.3Divorced51.4Spouse Died10.3Education levelIlliterate4612.8High School9025.0Diploma And Postgraduate Diploma15141.9Bachelor5615.6Master174.7job categoryChemical Storekeeper339.2Greenhouse Worker359.7Chemical Industry Worker20857.8Laboratory6317.5Mechanic215.8Educational status related to chemical safetyNo18952.5Yes17147.5Obtaining information related to chemical safetyOHS training courses10729.7Chemical safety label3810.6social media3710.3Self-Study133.6Multimedia (Specialized training and studying other resources)71.9No Info15843.9Enterprise sizeMicro-sized (< 10)308.3Small-sized (10–49)133.6Medium-sized (50–249)12233.9Larg-sized (250<)19554.2Chemical ExposureDaily22061.1Three Times Week4211.7Once Week6518.1Once Two Weeks267.2Once Month71.9 Demographic characteristics and work-related information of participants. Participants safety knowledge scores categorized based on enterprise size As described in the data collection tools, Sect. 2.4.1. Table 2 presents the frequency distribution of correct responses to multiple-choice questions about assessing participants’ chemical safety knowledge. In micro-enterprises, the lowest and highest awareness levels pertained to safe handling and storage of flammables (13.33%) and safe storage of reactive chemicals (100%), respectively. In medium and large enterprises, over 50% of individuals answered questions correctly. In large enterprises, the lowest awareness was associated with the meaning of the warning signal word (25.13%). Overall, participants answered over 70% of questions correctly, indicating acceptable chemical safety awareness. Furthermore, a comparison of participant responses based on enterprise size shows a statistically significant disparity in chemical knowledge between participants from large enterprises and those from micro, small and medium-sized enterprises in majority cases (P < 0.001). Table 2Distribution of the correct answers given for the MCQ questions of chemical safety knowledge evaluation Test.N0.multiple choice questionsCorrect AnswersP ***Micro (n = 30)Small (n = 13)Medium (n = 122)Big (n = 195)Total (n = 360)F* (n)P** (%)F (n)P (%)F (n)P (%)F (n)P (%)F (n)P (%)1Effects of harmful substances1136.67861.548267.2117087.1827175.27< 0.0012Definition of “Corrosive Substances”620.00861.549477.059548.7220358.39< 0.0013Definition of substances with “Acute Toxicity”723.33646.157460.666332.3115041.70< 0.0014Types of information on a label826.67646.159275.4118896.4129481.66< 0.0015The most reliable source of information on substance hazards (label and SDS, colleagues’ comment, …)1756.67861.548166.3917991.7928579.16< 0.0016Meaning of “Warning” signal word930.00323.08 ¥2923.774925.13 ¥9025 ¥0.8987Meaning of “Danger” signal word1653.33430.774436.077840.0014239.440.3408Meaning of “Caution & First Aid” phrase on the label1860.00646.156452.4616986.6725771.39< 0.0019Meaning of “may explode if heated” phrase on the label2376.67861.549477.0518695.3831186.39< 0.00110Solution to mismatch between gas cylinders color and its label2583.331184.629981.1519097.4432590.28< 0.00111Safe handling and storage of a pressurized cylinder826.67646.159174.5917991.7928478.89< 0.00112Proper action in the case of substance spill/splatter2686.671292.318670.4919097.4431487.22< 0.00113Mandatory use of safety goggles413.33753.854032.79 ¥8141.5413236.670.00714Basics of selecting an appropriate respiratory mask1653.331292.319275.4119298.4631286.67< 0.00115Types of information on the SDS2376.67753.858166.3918996.9230083.33< 0.00116safe handling of hazardous substances1653.33753.858267.2119097.4429581.94< 0.00117safe handling and storage of flammables413.33 ¥538.469376.2318594.8728779.72< 0.00118Best locations for installation of emergency eyewash and shower723.33646.157359.8416886.1525470.56< 0.00119safe transportation of hazardous substances1033.331184.629678.6918996.9230685.00< 0.00120safe storage of reactive chemicals30100.00 €13100.00 €9376.2318996.9232590.28< 0.00121Correct identification of the selected information (nameOf substance) on a GHS label2790.001184.628972.9518594.8731286.67< 0.00122Correct identification of the selected information(Signal word) on a GHS label2583.33861.5410283.61 €19197.95 €32690.56 €< 0.00123Meaning of the NFPA fir diamond symbol2893.331292.317662.309548.7221158.61< 0.00124Meaning of CAS number of chemical substances on the label1653.33861.548771.3111056.4122161.390.042*Frequency, **Percentage, ***Fishers exact test, € Maximum in each group, ¥ Minimum in each group. Distribution of the correct answers given for the MCQ questions of chemical safety knowledge evaluation Test. Correct identification of the selected information (name Of substance) on a GHS label Correct identification of the selected information (Signal word) on a GHS label *Frequency, **Percentage, ***Fishers exact test, € Maximum in each group, ¥ Minimum in each group. An assessment of the current state of chemical labeling in workplaces, based on 13 image-based items (as per Sect. 2.4.1), indicated that among participants from micro-enterprises, few answered correctly regarding oxidizing hazards (13.3%), while the highest correct responses pertained to flammability and toxicity and lethality of chemicals (80%). In small enterprises, 30.77% of participants demonstrated familiarity with reactive material hazards, whereas 100% recognized the “Toxic and Lethal” pictogram. In medium and large enterprises, the lowest familiarity levels were observed with the “Toxic and Infectious” pictogram (35.25%) and “Oxidizing Agent” pictogram (15.9%), respectively. Overall, 90% of participants recognized the “Chemical Explosion Hazard” pictogram, while the lowest familiarity rates were recorded for the “Oxidizing Agent” (25%) and “Toxic Substances” (39.44%) pictograms (Table 3). Furthermore, a comparison of chemical enterprises revealed statistically significant differences in the current status of chemical labeling practices across many of the hazard indicators (P < 0.05). Table 3Distribution of the correct answers given for the picture questions of chemical safety knowledge evaluation Test.N0.What is the meaning of this symbol?Correct AnswersMicro (n = 30)Small (n = 13)Medium (n = 122)Big (n = 195)Total (n = 360)F* (n)P** (%)F (n)P (%)F (n)P (%)F (n)P (%)F (n)P (%)P ***25 413.33 ¥753.854839.343115.90 ¥9025.00 ¥< 0.00126 2170.001292.319477.0517589.7430283.890.00227 2480.00 €13100.00 €9073.7718896.4131587.50> 0.00128 2376.671292.319779.5118896.4132088.89< 0.00129 2376.67969.239880.3318896.4131888.33< 0.00130 2480.00 €1292.319779.5119197.95 €32490.00 €0.89831 1446.671076.929073.779950.7721359.17< 0.00132 1446.67323.086049.186834.8714540.280.03633 1343.331292.3110182.79 €11156.9223765.83< 0.00134 1860.00861.549174.5910151.7921860.56< 0.00135 1343.33861.544335.25 ¥8543.5914941.390.21236 1653.33430.77 ¥7460.668945.6418350.830.02937 723.33930.005948.366734.3614239.440.003*Frequency, **Percentage, ***Fishers exact test, € Maximum in each group, ¥ Minimum in each group. Distribution of the correct answers given for the picture questions of chemical safety knowledge evaluation Test. *Frequency, **Percentage, ***Fishers exact test, € Maximum in each group, ¥ Minimum in each group. Participants safety awareness scores categorized based on enterprise size Participants’ familiarity and awareness of chemical safety standards and pictograms were significantly influenced by enterprise size. These findings were obtained using the GHS chemical labeling status assessment questionnaire (as per Sect. 2.4.2), which evaluated familiarity with labeling standards, practices, and pictogram recognition. Among participants, only 36.67% were familiar with the global GHS standard, with familiarity substantially higher in large enterprises (87.18%) compared to others (Q1; Table 4). However, 68.06% of respondents reported encountering unlabeled or damaged chemical packages (Q3 in Table 4), a challenge more prevalent in large (83.59%) and medium enterprises (51.64%). Conversely, reliance on personal experience (80%) as a primary source of safety information (Q7) and informal training methods (Q8) particularly in micro-enterprises (66.67%) was identified as a structural safety concern. While 45% of participants rated label positioning on packages as appropriate (Q6), only 41.11% assessed their ability to read labels as “good” (question 5). This deficiency was pronounced in small (69.23%) and micro-enterprises (66.67%), reflecting a linguistic barrier in accessing safety information. Significant differences in satisfaction with labeling systems were also observed: large enterprises (26.67%) reported the lowest satisfaction due to exposure to unfamiliar or multilingual labels (Q2). Table 4Distribution of the answers given for the familiarity and current status of chemical substance labeling evaluation Test.N0.QuestionsAnswersEnterprise categoryMicro (n = 30)Small (n = 13)Medium (n = 122)Big (n = 195)Total (n = 360)F* (n)P** (%)F (n)P (%)F (n)P (%)F (n)P (%)F (n)P (%)1Q1NO1963.33538.464032.792512.828924.72Yes1136.67861.548267.2117087.1827175.282Q2NO620.00215.386452.4614373.3321559.72Yes2480.001184.625847.545226.6714540.283Q3NO1653.33861.545948.363216.4111531.94Yes1446.67538.466351.6416383.5924568.064Q4NO2376.67753.859174.5918494.3630584.72Yes723.33646.153125.41115.645515.285Q5NO1033.33430.777158.2012765.1321258.89Yes2066.67969.235141.806834.8714841.116Q6NO620.00753.856250.8212363.0819855.00Yes2480.00646.156049.187236.9216245.007Q7NO620.001076.929174.5915881.0326573.61Yes2480.00323.083125.413718.979526.398Q8NO1550.00969.238468.8517288.2128077.78Yes1550.00430.773831.152311.798022.229Q9NO1033.33969.2310485.2517187.6929481.67Yes2066.67430.771814.752412.316618.3310Q10NO1033.331184.629376.2314674.8726072.22Yes2066.67215.382923.774925.1310027.7811Q11NO723.33969.238670.4915680.0025871.67Yes2376.67430.773629.513920.0010228.3312Q12NO826.67538.469477.0516484.1027175.28Yes2273.33861.542822.953115.908924.72 Distribution of the answers given for the familiarity and current status of chemical substance labeling evaluation Test. The assessment of pictogram familiarity, detailed in Table 5, revealed that the “toxic and lethal” (45.83%) and “flammable” (65.28%) pictograms were recognized as the most common symbols, whereas outdated symbols like “toxic” (WHMIS) were unfamiliar to 80.56% of participants. These results correspond to specific pictogram items in the questionnaire’s recognition section. Pictogram geometry also played a critical role in recognition: the GHS diamond (78.33%) and WHMIS circle (15.83%) were identified more accurately than the CHIP square (5.83%). Table 5Safety awareness and familiarity with pictograms among participation of each Enterprises.N0.What is the meaning of this symbol?AnswersEnterprise categoryMicro (n = 30)Small (n = 13)Medium (n = 122)Big (n = 195)Total(n = 360)F* (n)P** (%)F (n)P (%)F (n)P (%)F (n)P (%)F (n)P (%) 1 Never seen516.67323.084234.435829.7410830.00Only seen a few times310.00753.855746.725829.7412534.72seen a lot2273.33323.082318.857940.5112735.28 2 Never seen13.3300.004637.705930.2610629.44Only seen a few times1033.331292.316855.746935.3815944.17seen a lot1963.3317.6986.566734.369526.39 3 Never seen26.67215.382520.494523.087420.56Only seen a few times620.00753.855141.805729.2312133.61seen a lot2273.33430.774637.709347.6916545.83 4 Never seen723.33538.463327.054221.548724.17Only seen a few times620.00753.857158.206935.3815342.50seen a lot1756.6717.691814.758443.0812033.33 5 Never seen413.33215.382318.8521.03318.61Only seen a few times930.001184.626452.46105.139426.11seen a lot1756.6700.003528.6918393.8523565.28 6 Never seen1136.67646.154839.345829.7412334.17Only seen a few times13.33430.775645.906633.8512735.28seen a lot1860.00323.081814.757136.4111030.56 7 Never seen13.33538.464738.527035.9012334.17Only seen a few times826.67753.855040.988543.5915041.67seen a lot2170.0017.692520.494020.518724.17 8 Never seen2376.671076.927964.7517891.2829080.56Only seen a few times13.33323.084133.61147.185916.39seen a lot620.0000.0021.6431.54113.06 9 Never seen310.00430.777158.2015981.5423765.83Only seen a few times930.00861.544738.522311.798724.17seen a lot1860.0017.6943.28136.673610.00 10 Never seen2376.67538.465746.7217087.1825570.83Only seen a few times620.00861.545444.26199.748724.17seen a lot13.3300.00119.0263.08185.00 11 Never seen2583.33646.155948.369850.2618852.22Only seen a few times26.67430.774738.528342.5613637.78seen a lot310.00323.081613.11147.183610.00 12 Never seen930.001184.628670.4917489.2328077.78Only seen a few times1446.67215.383327.05189.236718.61seen a lot723.3300.0032.4631.54133.61 13 Never seen2170.00861.546351.6414875.9024066.67Only seen a few times620.00430.774637.702713.858323.06seen a lot310.0017.691310.662010.263710.28 14 What is the most geometric shape in which you have seen the pictograms within? 1653.33861.548771.3117187.6928278.33 826.67430.772520.492010.265715.83 620.0017.69108.2042.05215.83 15 What is the least geometric shape in which you have seen the pictograms within? 1240.00430.772218.0363.084412.22 516.67538.465645.907437.9514038.89 1343.33430.774436.0711558.9717648.89 Safety awareness and familiarity with pictograms among participation of each Enterprises. Participants safety Knowledge, safety behavior and risk perception scores categorized based on enterprise size Table 6 compares chemical safety knowledge, safety behaviors, and risk perception scores across enterprises of varying sizes. Findings reveal that chemical safety knowledge is significantly influenced by enterprise size (P < 0.001). Large enterprises exhibited the highest knowledge levels (18.97 ± 1.92), while micro (12.67 ± 2.40) and small enterprises (14.84 ± 3.72) scored lower. This disparity likely stems from large enterprises’ greater access to structured training programs, financial resources, and stricter regulatory compliance requirements. Table 6Knowledge, safety behavior and risk perception scores among participation of each Enterprises.VariableEnterprise categoryTotalMicro (n = 30)Small (n = 13)Medium (n = 122)Big (n = 195) P mean ± SDRangemean ± SDRangemean ± SDRangemean ± SDRangemean ± SDRangePicture Questions7.13 ± 1.704–119.15 ± 1.914–118.54 ± 2.601–128.10 ± 2.072–138.21 ± 2.261–13< 0.001Knowledge Scores19.80 ± 3.4913–2824.00 ± 5.0117–3124.39 ± 6.422–3327.08 ± 3.0517–3525.45 ± 5.031–35< 0.001 MCQ Questions 12.67 ± 2.409–1814.84 ± 3.7210–2015.85 ± 4.571–2218.97 ± 1.9210–2317.24 ± 3.771–23< 0.001Safety Compliance3.83 ± 0.722.75–5.753.56 ± 0.762.17–4.833.96 ± 0.622.25–5.253.69 ± 0.572.42–5.423.79 ± 0.622.17–5.170.001Safety Participation3.94 ± 0.962.55–5.553.60 ± 0.772.27–5.273.73 ± 0.702–53.62 ± 0.662.18–5.183.68 ± 0.712–50.104 Safety Behavior 3.88 ± 0.632.91–5.913.58 ± 0.522.7–4.653.85 ± 0.552.39–4.963.65 ± 0.532.74–5.743.74 ± 0.552.39–5.390.006Mystery Scale4.57 ± 0.723–53.69 ± 0.732.75–5.753.21 ± 0.811.75–5.754.05 ± 0.612–53.80 ± 0.821.75–5.75< 0.001Severity Scale3.27 ± 0.452.83–4.333.66 ± 0.363.17–4.333.61 ± 0.552.17–4.53.93 ± 0.382.33–4.673.76 ± 0.492.17–4.67< 0.001Hidden Health Effects Scale2.51 ± 1.151.20–4.43.72 ± 0.801.20–4.43.57 ± 0.781.4–5.44.05 ± 0.392.2–4.83.75 ± 0.781.2–5.2< 0.001Easy Spot Exposure Scale3.22 ± 0.322.33–4.332.76 ± 0.282.00–33.07 ± 0.831–53.77 ± 0.741–53.45 ± 0.811–5< 0.001Difficult Discriminate Symptoms Scale3.07 ± 0.932.33–5.333.94 ± 0.872.67–5.673.48 ± 0.991.67–5.673.73 ± 0.801.67–5.673.60 ± 0.901.67–5.67< 0.001Existence Risk Groups Scale3.28 ± 0.641.75–4.753.69 ± 0.941.75–5.753.87 ± 0.911.75–5.754.24 ± 0.571.5–5.54.01 ± 0.781.5–5.5< 0.001Mechanisms3.47 ± 1.061.00–4.6.00.63.89 ± 0.52.80–4.43.81 ± 0.741.4–5.43.75 ± 0.711.2–4.83.75 ± 0.751–50139 Risk Perception 3.33 ± 0.362.83–4.073.65 ± 0.332.77–4.773.55 ± 0.422.5–4.23.94 ± 0.262.97–4.473.75 ± 0.402.5–4.47< 0.001 Knowledge, safety behavior and risk perception scores among participation of each Enterprises. In contrast, safety behaviors showed smaller intergroup differences, with medium enterprises outperforming small ones (3.85 ± 0.55 vs. 3.58 ± 0.52). Overall safety behavior scores across all groups fell within the moderate-to-good range (3.68–3.88), suggesting employees’ relative adherence to safety protocols, even in small enterprises (P = 0.006). For risk perception, large enterprises reported the highest scores in perceived hazard severity (3.94 ± 0.26/out of 5 points) and hidden health effects (4.05 ± 0.39/out of 5 points). Conversely, micro enterprises scored highest on the Mystery Scale (4.57 ± 0.72), indicating potential ambiguities in safety information or training gaps. Additionally, large enterprises (3.77 ± 0.74) significantly outperformed small enterprises (2.76 ± 0.28) on the Easy Identification of Hazard Exposure Scale (P < 0.001), likely due to advanced monitoring systems and robust hazard reporting cultures in larger settings. Modeling and path analysis The direct paths between variables in the model used in the study and the standardized coefficients (β) are presented in Fig. 2; Table 7 The results indicated a statistically significant relationship between safety knowledge and safety awareness (P < 0.001), between safety awareness and risk perception (P < 0.001), and between safety knowledge and risk perception (P = 0.009). Additionally, the results of the direct relationships between variables in the model showed a statistically significant association between safety awareness and safety behavior (P < 0.001), safety knowledge and safety behavior (P < 0.001), and risk perception and safety behavior (P < 0.001). Conceptual model and tested research variables relationships. Table 7Direct paths and standardized regression Weights.PathsStandardizedeffect (β)Standarderror P Safety Knowledge → Safety Awareness0.6800.017< 0.001Safety Awareness → Risk Perception0.2810.012< 0.001Safety Knowledge → Risk Perception0.1820.0090.009Risk Perception → Safety Behavior0.2030.016< 0.001Safety Awareness → Safety Behavior0.2340.020< 0.001Safety Knowledge → Safety Behavior0.4340.014< 0.001 Direct paths and standardized regression Weights. As shown in Table 8, the proposed model demonstrates a good fit based on the fit indices. The results indicate that all the fit indices are within the acceptable range. Table 8Model fit indices.IndexValueAcceptance Modelχ2/df2.93< 3.00CFI0.96> 0.9NFI0.94> 0.9TLI0.92> 0.9RMSEA (90% CI)0.062 (0.051–0.076)< 0.08 Model fit indices. In the present study, the bootstrap method was employed to examine the indirect paths between variables and to determine the mediating effect of risk perception on the relationship between safety knowledge and safety behavior. A 95% confidence level was set, with 5,000 bootstrap resamples. As shown in the Table 9, the standardized beta coefficient for the mediating role of risk perception in the relationship between safety awareness and safety behavior is 0.0371, with P = 0.006. In other words, risk perception mediates the relationship between safety awareness and safety behavior. Table 9Indirect paths and standardized regression Weights.Indirect PathsIndirect EffectLower BoundUpper Bound P Safety Knowledge → Risk Perception → Safety Behavior0.2970.2190.379< 0.001Safety Awareness → Risk Perception → Safety Behavior0.0370.01540.07030.006 Indirect paths and standardized regression Weights. The standardized beta coefficient for the mediating role of risk perception in the relationship between safety knowledge and safety behavior is 0.297, with P < 0.001. This indicates that risk perception also mediates the relationship between safety knowledge and safety behavior. Discussion Chemical safety, as a fundamental pillars of occupational health, requires a deep understanding of hazards and risks, adherence to safety protocols, and access to educational resources across industries. A lack of sufficient knowledge and awareness of chemical risks can lead to serious accidents82. The GHS, as an international standard, plays a critical role in enhancing safety awareness and knowledge12,15. However, our findings, consistent with the theory of planned behavior, suggest that mere knowledge and familiarity of GHS labels and safety procedures is an insufficient predictor of safe behavior; it must be translated into a conscious motivation to act, often mediated by risk perception83. This study confirms the central hypothesis that enterprise size is a pivotal structural factor shaping employees’ chemical safety knowledge, preventive behaviors, and risk perception26,84. This finding resonates with the broader literature on sustainable manufacturing, which emphasizes how organizational resources and scale influence the implementation of safety and environmental practices85. The results confirmed significant variation across enterprise sizes (P < 0.001). The average chemical safety knowledge score in large enterprises was significantly higher than in micro and small enterprises. Moreover, over 50% of participants worked in large chemical enterprises with daily chemical exposure, while 40% lacked formal training due to employment in smaller enterprises. This gap can be attributed to several factors. Large enterprises often implement regular and structured training programs using resources such as SDS, educational videos and practical workshops. In these enterprises, access to online databases, available HSE consultants and tools for translating and learning pictograms and labels improve safety knowledge. In contrast, small and micro-enterprises rely mainly on informal training or personal experience of learning information from various sources29,77. Furthermore, supervision by national and international organizations requires larger enterprises to implement standards like GHS, whereas small and micro-enterprises often face legal exemptions or weaker oversight84. This disparity highlights a critical justice issue within occupational safety, where workers in smaller, often less-resourced enterprises are disproportionately exposed to risk26,86. When raw scores were converted to a 100-point scale, the average safety knowledge across all occupational and enterprise groups fell within the moderate range. This highlights two critical issues (1) even in large enterprises, knowledge did not reach the desired threshold (66.1–100); and (2) significant gaps persist in recognizing fewer common pictograms such as hidden health effects and mechanisms of injury. For example, the low familiarity with the WHMIS “toxic” symbol (80.56% unfamiliarity) underscores a specific, addressable gap in training content. Furthermore, 46.1% of participants in large enterprises reported accidental chemical exposure, underscoring the insufficiency of theoretical knowledge alone and the need for scenario-based practical training86. Safety behavior scores were significantly higher in medium and large enterprises compared to small ones (P = 0.006). In large enterprises, error-reporting cultures and safety participation are reinforced through reward-penalty systems, while micro and small enterprises often adopt reactive post-incident adjustments87. Moreover, advanced technologies such as gas leak sensors, closed circuit television (CCTV), and risk management software help automatically identify and correct unsafe behaviors in large enterprises88–90. However, the non-significant difference in safety participation scores across groups is a crucial finding (P = 0.104). It suggests that the willingness to voluntarily engage in safety initiatives is not inherently tied to enterprise size but is likely driven by factors like leadership, organizational culture, internal communication, and individual motivation, a point emphasized in studies on safety climate52,56,91. For instance, 84.62% of small enterprise respondents cited supervisor-led training as their primary learning method, highlighting leadership’s pivotal role. Risk perception was significantly higher in large enterprises (P < 0.001), particularly in subscales like severity, hidden health effects, and mystery. Employees in large enterprises perceived chemical risks as more severe, likely due to access to epidemiological data on long-term consequences (e.g., carcinogenicity) and enhanced training. Conversely, micro-enterprises scored highest on the mystery scale, reflecting unclear safety information and inadequate training, which aligns with the psychometric paradigm of risk perception, where unknown risks are often underestimated or misunderstood92. These findings align with realistic risk perception theory, which posits that larger organizations, with objective data and risk assessment systems, evaluate hazards more accurately, while smaller one’s underestimate risks due to subjective judgments34,37. In addition to enterprise size, other factors influence the outcomes. Regarding language and information access, results indicated that only 40% of participants rated their ability to read labels and communicate clearly as good. This issue is more critical in small enterprises, which receive fewer localized (Farsi) labels. Furthermore, implementing safety standards such as ventilation systems or protective gear is costly, and small enterprises, particularly in developing countries, often lack the financial capacity to afford them. In micro-enterprises, prioritizing production and profitability may overshadow safety concerns93. The core of our analysis, addressing the study’s primary objective, revolves around the mediation model. The SEM analysis exhibited that safety knowledge and awareness not only directly influence safety behaviors but also exert a significant partial mediation effect through heightened risk perception (β = 0.203, P < 0.001). This provides robust empirical support for the theoretical mechanism suggested by the reviewer: employees with greater safety knowledge and GHS awareness are more likely to understand the potential consequences of chemical hazards such as injury severity or hidden health effects and this deeper cognitive understanding of risk motivates stricter adherence to safety protocols. These findings align with Protection Motivation Theory, which posits that risk appraisal is a key mediator between motivational mechanisms (e.g., fear of consequences) and the adoption of protective measures as safety behavior35. The results suggest that training programs should move beyond knowledge transfer to include realistic hazard simulations and objective risk perception enhancement to foster lasting behavioral improvements29,91. Limitations and future study While this study holds significant potential, several limitations should be noted. The cross-sectional design prevents inferring causality. The focus on chemical industries in a specific developing country like Iran may limit the generalizability of findings to other nations or sectors. Furthermore, self-reported data may be influenced by social desirability bias. Additionally, reported scores may be influenced by respondents’ tendency to present themselves favorably, introducing potential social desirability bias. The study also did not explore the role of individual factors such as gender or work experience as moderating variables in safety behavior analyses. To reduce disparities between enterprises, strategies such as targeted training programs, standardization and oversight, and the adoption of assistive technologies can be proposed. Developing free, visually oriented training courses (e.g., virtual reality videos) for small enterprises and integrating chemical safety education into technical and vocational training curricula can enhance knowledge and awareness. Establishing a national chemical incident reporting system to identify hazard patterns, designing mobile applications for real-time label translation and emergency action guidance, and deploying AI-driven hazard prediction systems based on historical data will integrate information and enrich chemical safety protocols across all types of enterprises. Future research should employ longitudinal designs to establish causal pathways and explore the moderating role of individual factors like gender and work experience and evaluating the impact of training programs on improving safety knowledge. A comparative analysis between high- and low-income countries could further clarify the role of national infrastructure and policy in shaping chemical safety practices. Conclusion This study demonstrated that, in addition to individual and environmental characteristics, the size of chemical enterprises as a critical structural factor significantly influencing chemical safety knowledge, GHS awareness, safety behaviors, and risk perception. Several key findings emerge from our analysis. first, there are substantial disparities in safety competence across enterprises, with large enterprises showing significantly higher scores in safety knowledge (27.08 ± 3.05) and risk perception (3.94 ± 0.26) compared to micro-enterprises (19.80 ± 3.49 and 3.33 ± 0.36 respectively). Second, we identified a critical gap in specialized knowledge, evidenced by the 80.56% unfamiliarity with toxic material pictograms and the heavy reliance on informal training in small enterprises (84.62%). The major contribution of this research is the empirical validation of a mediation model, identifying risk perception as the psychological mechanism that translates safety knowledge and awareness into improved safety behaviors (β = 0.203, P < 0.001). This finding provides a mechanistic explanation for how knowledge transforms into action in chemical safety contexts. The structural advantages of larger enterprises including access to structured chemical safety training resources, stricter regulatory oversight, and advanced monitoring systems create an environment conducive to safety excellence. In contrast, small and micro-enterprises face compounded challenges stemming from inadequate organizational training, reliance on informal education, and financial constraints that limit safety investments. These findings indicate that improving chemical safety requires an integrated approach that simultaneously focuses on enhancing knowledge and awareness, strengthening risk perception, and facilitating access to resources. Furthermore, the identification of risk perception as a mediator in the relationship between safety knowledge/awareness and safety behaviors underscores the importance of designing training programs based on realistic hazard simulations. Bridging the existing gap necessitates tripartite collaboration among governments (policy formulation and subsidies for SMEs), industries (standard implementation and knowledge sharing), and academia (development of assistive technologies and training methodologies). Only through such synergistic, evidence-based approaches can safer and more sustainable workplaces be achieved across the chemical industry spectrum. Practical implications and contributions Based on the empirical findings of this study, several practical contributions can be outlined across four key strategic domains to bridge the safety gap between enterprises of different scales. These recommendations directly address the specific deficiencies identified in our results, such as the unfamiliarity with certain pictograms and the mediating role of risk perception. First, regarding targeted training and capacity building, it is imperative to design tailored programs for SMEs that move beyond theoretical knowledge. As our findings showed that 80.56% of participants were unfamiliar with the toxic materials pictogram, training must use visual content like short videos and infographics, with a dedicated focus on these less common but critical pictograms (e.g., toxic and flammable). Furthermore, to directly strengthen the identified mediator of risk perception, governmental authorities or industry associations should organize interactive workshops featuring realistic hazard simulations, making the potential consequences of unsafe behavior more tangible. Second, standardizing labels and improving information accessibility is crucial to address the linguistic barriers we identified. Mandating bilingual labels (Farsi-English) on all chemical products and establishing a national, freely accessible Farsi-language SDS database would significantly increase informational transparency and empower workers. Third, integrating technology into safety management can provide scalable solutions. Developing mobile applications for real-time label translation and emergency action guidance can be a low-cost tool for SMEs. For larger enterprises, deploying IoT sensors for real-time chemical leak monitoring can automate hazard detection and reinforce safety behaviors. Fourth, implementing supportive and incentive-driven policies is essential to motivate compliance. This includes providing financial subsidies to SMEs for acquiring safety equipment and granting tax exemptions to enterprises that achieve international safety certifications (e.g., ISO 45001) or demonstrate full GHS compliance. Creating a public safety rating system for enterprises would enhance transparency, foster competitiveness, and ensure continuous improvement. These evidence-based strategies not only reduce disparities between small and large enterprises but also improve safety behaviors across the chemical industry by strengthening risk perception as a mediating mechanism. Successful implementation requires a tripartite collaboration among governmental bodies (policy and funding), industries (implementation and knowledge sharing), and academic institutions (technology development and evaluation) to foster safer and more sustainable workplaces. Acknowledgements This study is related to the project NO. 1402/5366 From Student Research Committee, Shahid Beheshti University of Medical Sciences, Tehran, Iran. We also appreciate the Student Research Committee and Research & Technology Chancellor in Shahid Beheshti University of Medical Sciences for their financial support of this study. Author contributions SV.E: Conceptualization, Project administration, Writing – review & editing, Methodology, Investigation, Visualization. R.E: Writing – original draft, Methodology, Investigation. A.M and M.B: Data curation, Writing – original draft. M.J: Formal analysis, Data curation, Writing – review & editing. A.K: Formal analysis, Methodology, Resources. Investigation. M.P: Conceptualization, Project administration, Investigation, Writing – review & editing, Supervision, Resources. All authors review and approved the final manuscript. Funding This study was part of the research projects supported by Student Research Committee, Shahid Beheshti University of Medical Sciences (Grant no. 43007311; 1402/5366). Data availability Data supporting this study are included within the article and/or supporting materials. Declarations Competing interests The authors declare no competing interests. Conflict of interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Ethics approval This study was performed following with the principles of the Declaration of Helsinki and was approved by the Ethics Committee of Shahid Beheshti University of Medical Sciences (IR.SBMU.RETECH.REC.1402.467). Moreover, all participants provided written informed consent before participating in the study. The informed consent form included elements such as the introduction of the research, the procedures involved, the manner of participation, benefits and potential side effects, costs, alternative methods, confidentiality of information, the researcher’s accountability to answer queries, the right to decline or withdraw from the study, and the affirmation of the consent form. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement Data supporting this study are included within the article and/or supporting materials.