Biases in the Safety and Security Risk Management of Chemical-Related Academic Laboratories

Abstract

Based on the interpretation of the identified risk biases from a narrative perspective, this paper studies the biases in safety and security risk management of chemical-related academic laboratories from four parts: risk identification, risk assessment, risk control and continuous monitoring. Mainly systematic error, inclusion of risk events, cognitive factors, model/algorithmic and social/interpersonal during risk management are discussed. The bias related to uncertain risk events, which is the most common and easily ignored during risk management, mainly including the imbalance between safety risk management and security risk management. Therefore, while protecting the laboratory from unintentional and unpremeditated safety risks within the system, it is also critical to protect the system from external, deliberate and premeditated security risks. This research paper is expected to spur and promote more discussion and the best practices in laboratory risk management among researchers, educators, managers and other stakeholders for handling biases in the risk management of chemical-related academic laboratories.

1. Introduction

Academic laboratories are the central focus of safety management in colleges and universities due to their interdisciplinarity, intensive use of instruments and equipment and the large number of students involved [1,2]. Especially for chemical-related academic laboratories, the heightened safety and security risks are characterized by (1) a diverse array of hazardous chemicals and the regular use of equipment under extreme temperature and pressure; (2) an experimental process, which is exploratory and fraught with unknown risks, including unpredictable interactions of potential hazards; (3) a constant turnover with seasoned researchers graduating and new members joining annually; and (4) innovation, which is a core aspect of scientific research, and the resources within laboratories, including personnel, data and equipment, which are highly valuable. Therefore, much attention should be paid to the safety and security risks of chemical-related academic laboratories [3].

Laboratory safety refers to the protection of the system from internal, unintentional, unpremeditated actions or factors resulting in safety incidents (e.g., chemical spills, laboratory explosions, etc.). Laboratory security refers to the protection of the system from external, intentional, premeditated actions or factors resulting in security incidents (e.g., chemical theft, terrorist attacks against chemical and process equipment, intelligence theft, etc.). Major risks exist if there are failures in control within the context of chemical laboratories. The integrated development of Safety and Security is still the main problem of risk management in chemical-related academic laboratories [4]. Risk management is the key procedure to ensure laboratory safety and security in colleges and universities. The process can be summarized as risk identification, risk assessment, risk control and continuous monitoring, which is expected to be an effective way to identify, assess and control potential threats in the laboratory. Nevertheless, safety and security incidents still occur worldwide occasionally [1]. Among all laboratory types, chemical laboratories have the highest number of accidents [5]. For example, in 2008, a tert-butyllithium accident in an academic laboratory at University of California, Los Angeles (UCLA) resulted in the death of a research assistant [6]. In 2018, an explosion at Beijing Jiaotong University resulted in the death of two doctoral and one master’s students [7]. In 2013, a graduate student of Fudan University stole the highly toxic chemical N-dimethylnitrosamine from the laboratory and killed his roommate by poisoning. These incidents expose the defects and challenges in the safety and security management of chemical-related academic laboratories. In addition, the number of incidents in chemical-related academic laboratories is significantly higher than that in industrial laboratories [8] due to the existence of biases, which may lead to distorted risk descriptions, incomplete risk analysis or unexpected disasters [9,10]. Therefore, this paper aims to provide ideas for the bias during the identification, assessment, control and monitoring of safety and security risks in chemical-related academic laboratories in universities to provide theoretical support and practical guidance for the risk management strategy.

2. Methodological Approach

This paper can be viewed as conceptual research on the risk biases discussed, covering systematic bias, bias associated with uncertain risk events, cognitive bias, etc. [11]. The paper adopts a narrative perspective, and reasoning and argumentation are the key instruments. This type of research work builds on a review of the current situation and deficiencies of risk management in chemical-related academic laboratories. The following examples illustrate these elements for the present study:

• Risk identification: The types of biases in risk management that may exist in the process of hazard source identification in chemical-related academic laboratories were discussed, including the biases in the identification processes of safety risks and security risks.
• Risk assessment: It shows the biases existing in the currently widely used risk assessment methods. Considering the integration process of safety and security implemented in various industries nowadays, it focuses on discussing the imbalance of safety risks and security risks in risk assessment.
• Risk control: The possible risk biases in policy making, safety and security assurance and cultural promotion in chemical-related academic laboratories are discussed.
• Continuous monitoring: Reasoning and demonstrating the possible biases and deficiencies in the process of continuous monitoring of risk management.

3. Making Sense of Risk Biases

Bias is a persistent problem that all scientific disciplines grapple with, and risk science is no exception. Risk research often involves assumptions, data and information collection, qualitative and quantitative analysis, review and judgment, decision-making and communication. Thus, bias can arise in data collection, analysis, interpretation, communication and perception, among others [11]. Recently, Thekdi et al. discussed four types of biases (systematic error, risk event, model and cognitive factor) that can directly affect risk research by exploring the characteristics of currently identified biases (cognitive biases, behavioral biases, algorithmic biases and social/interpersonal biases) [11]. Based on the results of Thekdi et al., considering the characteristics of chemical-related academic laboratories and the current situation of risk management, this section discusses the biases of risk management of chemical-related academic laboratories in universities as follows.

• Systematic bias: The investigation process of risk research and risk management in university academic laboratories inevitably involves data collection. Systematic bias refers to systematic errors in the measurement of risk research due to its specificity, mainly involving potential biases in the data collection process. For example, a laboratory classification method based on one cultural context may not be applicable to laboratories in all cultural contexts.
• Bias associated with uncertain risk events: In general, risk is conceptualized as a combination of uncertainty and consequence, which is denoted by (U, C). Information regarding risks is inherently incomplete, and the knowledge of risk analysts, risk managers and other stakeholders is limited. Consequently, it is not possible to precisely measure all potential states of future events and the probabilities of their occurrence. In the process of risk assessment or risk decision-making, the exclusion of significant risk factors or events leads to inaccuracies and incompleteness in risk practices. Given this, bias associated with uncertain risk events can be understood as biases that are required to be addressed within the objectives and scope of risk research. However, these biases are often overlooked in actual risk research or risk practice [11]. The reasons may include intentional research design, deviations between theory and practice or methodological deficiencies. Near misses or “Black Swan” incidents can be used as examples of biases related to uncertain risk events. In chemical-related academic laboratories, it mainly involves the comprehensive management of safety and security aspects that have not been taken into account in risk management.
• Cognitive bias: In the process of risk identification, assessment, decision-making and continuous monitoring, individuals’ differences in the perception of events, information and knowledge usually result in personal errors in risk handling and judgment. This phenomenon, which leads to biases in aspects such as risk factors, safety and security events, and development trends due to individual subjective tendencies, affective preferences or limited information can be understood as cognitive biases. According to Tversky et al., cognitive biases often affect the judgment of probability and uncertainty [12,13]. In risk management of chemical-related academic laboratories, cognitive biases can be summarized as representativeness bias, availability bias and anchoring bias, among others. Representativeness bias refers to the bias in the judgment caused by people’s excessive attention to certain representative features. Managers may rely on a small number of non-random samples or limited personal experience when making decisions, which cannot guarantee the integrity of the information obtained in the process of risk judgment [13]. Availability bias denotes the propensity for individuals to assess the likelihood of events based on their cognitive availability, that is, relying on certain information or occurrences that are easier to recall. Managers are easily dominated by first impressions or first information when making judgments about risk events. Anchoring bias refers to the bias caused by people’s tendency to use previously obtained information as a reference for processing or judgment, even if that information is inaccurate.
• Algorithm/model bias: In risk identification, assessment, control or other algorithm/model-based applications, some biases, due to assumptions, data collection and analysis, model fitting and other factors, are considered algorithm/model biases in risk management. For example, the assumptions of the model and the selection of variables are not suitable for academic laboratories, which may lead to biased results of the model. Furthermore, the data used to train the model are not representative of the entire target group, which can lead to inaccurate predictions of underrepresented groups.
• Social/interpersonal bias: In team or group settings, interpersonal bias can lead to false consensus and group-thinking behaviors, which can compromise the multi-stakeholder and multi-standard nature of complex risk issues [11]. At the societal level, both explicit (directly expressed) and implicit (indirectly implied) biases can have serious implications for equity in the context of risk, such as the overly broad consensus in laboratory safety and security inspections.

Several categories of biases that may affect risk management in chemical-related academic laboratories have been discussed above. It is important to note that these biases are not viewed in isolation and that there are potential links between different categories. For example, due to the availability bias in cognitive factors, researchers or managers may judge the likelihood of an event based on their cognitive availability, leading to the neglect of certain risk factors or events. Therefore, the exploration of biases in risk management cannot be isolated, and risk management should be systematically combined with risk biases to ensure the completeness and accuracy of the analysis and judgment.

4. Analysis of Notable Biases in Risk Management of Academic Laboratories

In its broadest conception, risk can be conceptualized as the combination of uncertainty and consequence [14]. Uncertainty, as a principal aspect of risk, has not yet been adequately addressed, with potential and unforeseen issues remaining unresolved to date. The characteristics of risk determine that no matter what preventive measures are taken, there is always uncertainty about the occurrence of future incidents [15]. Given this, risk management is a continuous process that requires regular monitoring and assessment to address emerging risks and changing conditions. It also requires laboratory managers and decision makers to consider potential risks and consequences. Therefore, based on the bias categories of risk science research that have been identified, this study, through critical thinking of the past and present risk management of chemical-related academic laboratories, provides a detailed explanation of the biases in risk management (i.e., risk identification, risk assessment, risk control and continuous monitoring). Based on the current situation of laboratory risk management, Figure 1 shows the overall framework of risk management biases in chemical-related academic laboratories.

4.1. Risk Identification

Risk identification is the first step in laboratory risk management, which includes a systematic review of all potential hazards/threats in the laboratory environment, which may include chemical spills, equipment failures, biological risks, etc. [16]. In laboratory settings, hazards/threats refer to unsafe human behaviors, hazardous states of objects and managerial deficiencies (that is, the conditions of energy, substances and actions, rather than the energy, substances and actions themselves). Risk, as the combination of the likelihood of an event occurring and its consequences, may originate from both internal (safety incident) and external (security incident) factors of the system. The identification of hazards in chemical-related academic laboratories is the basis of risk assessment. Only comprehensive and systematic identification of risk sources in academic laboratories can ensure the effectiveness of subsequent risk assessment and control.

Over the past few decades, many different methods of risk identification have been developed. These methods include, for example, Safety Checklist (SCL), Preliminary Hazard Analysis (PHA), Hazard and Operability Study (HAZOP), Fault Modes and Effect Analysis (FMEA) [17]. Many methods have been applied in the process of risk identification in chemical-related academic laboratories. For instance, Peng et al. used a safety checklist to collect safety data from Southwest University laboratories for further risk analysis [18]. At present, safety risks associated with chemistry, physics, electricity, biology, mechanics and psycho/physiology are the primary sources of hazards and threats discussed in laboratory risk research. For example, chemical hazards include explosions caused by the incorrect use of hazardous chemicals, gas leakage such as methane [19,20]; physical hazards involve cutting, slipping/falling of broken glass instruments [21]; electrical hazards include wire short circuit and electrostatic discharge [19]; biological risks involve infections related to bacteria and microorganisms [22]; mechanical hazards include incidents caused by the use of mixers or grinders [23]; and psychological hazards involve fatigue and ergonomic workload.

In the process of risk identification in chemical-related academic laboratories, there may be biases related to cognitive factors. Firstly, risk identification depends on the knowledge and experience of risk analysts, which leads to the subjectivity of risk identification. That is, the hazard sources of the same system are identified by different people and the results may be quite different. Concurrently, different risk managers and stakeholders may have different perspectives, opinions or tendencies. This divergence in thinking can lead to situations where properly conducted risk identification is interpreted as misinformation, which in turn leads to doubts about the credibility of risk research and the effectiveness of risk management [24]. Consequently, the formation of professional, reliable and comprehensive risk analysis teams may reduce the subjectivity of risk identification. As a reference for risk management in academic laboratories, the establishment of interdisciplinary teams may be effective. For example, Marroni et al. assessed the attractiveness of chemical and process facilities to terrorism by assembling interdisciplinary teams of experts, including those in geopolitics, security and chemical processes [25].

In contrast to laboratory safety risks, laboratory security risks have not been widely discussed. It is worth noting that a system may have both safety-related and security-related risks, and both may adversely affect the system (e.g., personnel casualties or property losses) [26]. Therefore, risk management in chemical-related academic laboratories should protect humans from chemical hazards while also protecting chemicals from human misconduct [27].

4.2. Risk Assessment

Risk assessment is the process of analyzing the uncertainty and severity of identified risks through qualitative and quantitative methods and providing information for managers and decision makers. Reviewing the traditional research of safety science, the risk factors of chemical-related academic laboratories in colleges and universities mostly involve human factors, materials and equipment, environment and management. The process of assessing risk factors may involve data collection. Questionnaires, as a practical method for assessing laboratory safety risks, rely on the accuracy of respondents’ self-reported data, which have inherent biases (e.g., social expectation biases). These biases include, for example, the results of safety behavior in the academic laboratory where the respondents share their opinions (social/interpersonal bias) or are influenced by the wording or order of questions (systematic bias) [23]. In terms of qualitative research, an interview is expected to generate knowledge based on human experience in order to obtain insights by mining problems and understanding/analyzing human behavior and psychology [28]. The interview process may have biases from both the interviewers and the interviewees. For example, the interviewer’s attitude, preconceived views and misunderstanding of the interviewees’ answers may affect the accuracy of information collection. Concurrently, it is difficult to achieve complete neutrality in the interview process due to interpersonal interaction. In addition, other qualitative and quantitative methods are also widely used in laboratory safety risk assessments, including the Fuzzy Analytic Hierarchy Process (FAHP), Human Factor Analysis and Classification System (HFACS-UL) and the Bayesian Network, among others [20,29,30]. It is worth noting that a single risk assessment method may not be sufficient to achieve high-quality risk analysis. Therefore, the combination of multiple methods is expected to make the risk assessment more accurate [17]. In most cases, the choice of risk assessment methods is based on the personal preference and experience of the researcher, which can lead to cognitive biases. Furthermore, most investigations are confined to samples from a specific cultural background, resulting in findings that are only applicable to a limited scope. In the future, more cross-cultural investigations are expected to reduce representativeness bias in laboratory risk studies. It is necessary to emphasize that systematic biases (potential biases during data collection) of risk studies in academic laboratories are inherent. Self-reported data, for example, have inherent limitations. Nevertheless, risk researchers must acknowledge and identify the limitations of their research in the process of data and information collection to show that their findings are applicable to specific cultural contexts and groups.

Risks are inherently dynamic, and events that once had a low probability of occurrence can potentially change over time due to factors such as system updates and personnel changes. Therefore, risk identification and risk assessment should not be regarded as one-time tasks. In contrast, regular risk identification and risk assessment can make laboratory operators, managers or other stakeholders aware of the existence of risks timely and accurately so that measures and actions can be taken to reduce the probability of near misses, incidents and accidents. In risk assessment, probability is often used to represent uncertainty. At the same time, the strength of knowledge must be used as an important factor to support probability judgment when assessing event probability [31]. Therefore, the accuracy of risk assessment also depends on the strength of knowledge. Aven and Flage et al. pointed out that “the reasonability of assumptions”, “the reliability of data/information”, “the consistency of experts” and “the understanding of phenomena” are critical factors in evaluating the strength of knowledge [15,31]. In the risk analysis of chemical-related academic laboratories, unreasonable assumptions may cause the analysis results to deviate from the real situation, thus affecting the effectiveness of risk control and management decisions. Reliable data and information are the basis for effective risk assessment. As mentioned earlier, inherent biases in data collection and unreliable data (e.g., outdated, inaccurate or incomplete) can lead to inaccurate assessment results. The consistency among experts can significantly enhance the credibility and authority of the results. In addition, the expert’s complacency can also lead to bias in risk analysis [32]. Therefore, to obtain a broad consensus of experts and reduce the biases caused by cognitive factors, it is crucial to concurrently avoid the social/interpersonal bias that may stem from group thinking. In recent years, there has been growing recognition of the limitations of traditional risk assessment methods in capturing the potential for unpredictable events and risk uncertainty [33,34,35]. Therefore, under the background of risk analysis and risk management, the strength and dynamics of knowledge play an important role in reasonable assumptions, reliable data, expert consensus and sufficient understanding of the phenomenon/system. The emergence of new information and data about systems or phenomena may help increase the predictability of risk events [36].

Resources for controlling safety and security risks at most institutions are limited [37]. It is important to implement comprehensive management so that resources are used effectively. Thus, safety and security science research should be intertwined and inseparable [38]. However, practitioners have traditionally viewed safety and security as distinct attributes, often working in isolation using their respective frameworks [39]. Risk research in scientific production is no exception: the current risk assessment and practice of chemical-related academic laboratories mostly regard safety and security issues in isolation and cannot fully understand and identify the threats from the internal and external systems. For example, Zhao et al. constructed a Bayesian network model by calculating the conditional probabilities of four factors, namely human, equipment and material, environment and management, to identify the relative importance of risk factors that lead to university laboratory safety incidents [40]. Liu et al. established a risk assessment method based on the analysis of the causes of accidents in chemical laboratories in colleges and universities to identify the sources of danger in chemistry laboratories and to determine the priorities for risk management [41]. In other words, the current risk assessments of academic laboratories predominantly focus on safety risks, with only a minority of studies addressing security risks. It is undeniable that isolated research on safety and security risk analysis will lead to the limitations of risk management in chemical-related academic laboratories. As an example, considering only safety risks, an open laboratory door may facilitate evacuation procedures in case of fire or explosion. However, an open door would undoubtedly elevate security risks, potentially leading to the loss of experimental data, destruction of laboratory equipment and misuse of chemicals, among other laboratory security incidents [42].

Considering the imbalance between safety risk analysis and security risk analysis in chemical-related academic laboratories, the EVIL DONE attribute analysis based on the Situational Crime Prevention (SCP) principle can be used as a practical method to identify and assess security risks in academic laboratories. Clarke et al., based on the Situational Crime Prevention principle, developed eight attributes to describe the motive and opportunity of a crime, namely EVIL DONE, which is acronyms of Exposed, Vital, Iconic, Legitimate, Destructible, Occupied, Near and Easy [43]. Marroni et al. used the EVIL DONE attribute to capture all potential drivers that could motivate terrorists to launch physical attacks on chemical and process equipment, aiming to assess the attractiveness of these pieces of equipment to terrorism [25]. To adapt the EVIL DONE attributes for security risk analysis in chemical-related academic laboratories,

Table 1. EVIL DONE attributes and descriptions for chemical-related academic laboratories [43].

Attribute	Description	Example
EXPOSED	The characteristics of chemicals, chemical equipment or critical data/information are visible and can attract attention.	Unlocked chemical cabinets or unsupervised equipment are more exposed.
VITAL	The tendency for academic laboratories to play a key role within a specialized field or for chemicals, equipment/instrument, etc., to play a key role in exploratory experiments.	The importance of State Key Laboratory is significantly higher than ordinary laboratory.
ICONIC	The target has the characteristics of the symbolic value.	Compared with ordinary universities, the academic laboratories of military-related universities are more vulnerable to malicious attacks such as intelligence theft.
LEGITIMATE	Target that elicited a positive response from the public.	Laboratories that cause long-term environmental pollution may be more vulnerable than sustainable laboratories.
DESTRUCTIBLE	A device is completely destroyed or severely damaged by an attack using as few resources as possible and as simple a tool as possible.	Targets that can be destroyed with simple tools are more vulnerable to malicious attacks than targets that require explosives or more advanced means to destroy.
OCCUPIED	Targets that can involve more casualties or more serious consequences.	Most security incidents, such as poisoning, are the theft of more toxic and corrosive dangerous chemicals.
NEAR	Targets that are close to the attacker.	Compared to targets that require prolonged routes, attackers may choose a laboratory that is close to the place of residence as a target.
EASY	Characteristics of physical security measures that are more vulnerable to breach.	Laboratories lacking surveillance systems, alarm systems, or regular patrols by security personnel are more susceptible to malicious security incidents such as the theft of chemicals and experimental data.

presents adjustments to the EVIL DONE attributes. For example, proper storage of hazardous chemicals, dual-lock systems for pharmaceutical cabinets and ensuring supervision during equipment operation are sound safety and security practices that can significantly mitigate “EXPOSED” risks.

Taking into account attractive factors from within or outside the academic laboratory, the likelihood of an information security incident is higher when the laboratory’s computers or databases store critical information/data but have weak access controls; this probability may increase when the experimenter leaves the laboratory with the computer still on or without a password. Furthermore, when chemicals that are flammable, explosive or could be used as precursors for drugs or weapons are exposed to the public view, there are significant risks of laboratory safety and physical security incidents. This probability may increase when the physical security systems of the laboratory are inadequate. The superposition of various factors, especially the insufficient security system of chemical-related academic laboratories and the lack of attention of laboratory personnel to security risks, makes the risk management of university laboratories challenging [18].

4.3. Risk Control

The risk management and control of chemical-related academic laboratories in colleges and universities involve the factors of policy making, safety and security assurance and cultural promotion. On the one hand, it aims to take strategies to gain opportunities to promote organizational development and, on the other hand, to avoid losses, incidents and catastrophes. Risk policies in the laboratory delineate the objectives, methodologies and procedures required to achieve laboratory safety and security, which may encompass the establishment of documentation, regulations and standards. In China, for example, “Safety Standards for Higher Education Laboratory”, a document of the Ministry of Education, guaranteed safety in laboratories. In addition, colleges and universities have the same or different policies for laboratory safety. Most of these policies are laboratory risk management principles that only express normative obligations in certain scenarios, rather than prescribing specific—compliance and practice depend on specific conditions and situations [26]. Therefore, there may be a gap between management’s stipulation of objectives/procedures and specific implementation in chemical-related academic laboratories. This can be explained as the gap between theory and practice. Aven pointed out that the four types of gaps between risk theory and risk practice are (1) the transformation and diffusion of risk knowledge from theory to practice, (2) the objective of risk science theory and practice, (3) the identification of two types of risk theory and (4) the resistance psychology of change management [14]. In the field of laboratory safety and security, this gap also manifests as an issue of cognitive bias in risk research and management. For example, if the laboratory staff had a misunderstanding of safety and security objectives and procedures, the feedback and actions would be delayed. The adequacy in the theoretical knowledge concerning risk science among risk analysts and policymakers and the frontier of application are also significant factors. This leads to question whether laboratory safety and security standards are up to date. Furthermore, the focus on the laboratory security policy and management is not enough. The worldwide reported incidents of chemical theft and attack in universities have sounded alarms about security risks [44].

Safety and security assurance in chemical-related academic laboratories refers to the actual work carried out according to the policies, norms and standards established by management. Despite significant differences in some aspects of safety and security, they show great similarity in the comprehension and mitigation of the risks in both domains [38]. It involves specific management processes such as laboratory inspections, emergency response mechanisms, risk minimization and ensuring safety and security facilities. For example, during the experiment, the complexity of laboratory access should be reduced and the escape and emergency passage should be unimpeded. Laboratories should be promptly secured when unattended to prevent unauthorized intrusion by malefactors. These measures not only reduce the likelihood of safety incidents but also help prevent the occurrence of security incidents. However, due to availability bias, managers or experimentalists pay too much attention to the laboratory hazard sources they already know or easily understand, thus ignoring the in-depth exploration of other risk information, resulting in errors in judgment, communication and decision-making. Concurrently, influenced by the anchoring effect, the laboratory staff’s previous experiences with near misses, incidents and accidents may affect their perception of laboratory risks, leading to a bias in risk perception. According to Qiu et al., workers who have experienced minor accidents may adopt a casual attitude toward unsafe behaviors, thus reducing their risk perception [45]. However, as the severity of the accident increases, the severe impact can amplify their recognition of potential dangers, significantly elevating their risk perception of unsafe behaviors. Furthermore, top-down management provides a means of safety and security, while bottom-up feedback provides early warnings of residual hazards [46]. An investigation of academic institutions found that although some laws and regulations on chemical safety and security exist in the Philippines, effective enforcement systems are often lacking because none of the participating institutions had relevant plans or programs [47]. There is no doubt that this top-down implementation problem, which is prevalent in higher education institutions, increases the safety and security risks of research and production.

The safety and security concept of different stakeholders may lead to the limitation of risk management in academic laboratories. Cultural promotion in academic laboratories refers to laboratory safety and security education and training, information communication and all related activities necessary to maintain a positive safety and security culture. Individuals’ risk perception and ability to detect potential hazards can be markedly improved through the acquisition of new knowledge, exposure to innovative concepts and the mastery of new techniques [48]. It is worth noting that the education and training received by laboratory safety managers and graduate instructors in universities and courses have been practiced and promoted for many years. Their belief and pride in established thinking and methods has led to resistance to change [14]. This will lead to a cognitive bias in risk research. In addition, although universities and scientific research institutions are actively creating a good safety atmosphere to improve students’ safety compliance and safety participation, unsafe behaviors still exist in academic laboratories [49]. It is inevitable that different stakeholders disagree when considering laboratory risks, which may cause friction between organizations. For example, Adawe showed that the effectiveness of chemical safety and security systems hinges on the awareness and vigilance of the practitioners involved [50]. Currently, in most universities in China, students receive risk and safety knowledge through lectures and course guidance [51]. Virtual laboratories based on Web technology and VR technology can be used as an effective training strategy to improve the safety and security knowledge and awareness of chemistry students, researchers and managers [52]. This virtual teaching of risk scenarios can help experimentalists simulate behavior under different conditions where their behavioral responses can be observed and corrected, improving experimentalists’ safety and security skills. While it is a large investment for universities and some stakeholders, it is also an effective way to create a positive safety and security culture and raise students’ awareness of safety and security risks for laboratory safety and security managers. Furthermore, the prioritization of scientific research production and laboratory safety and security is worth considering for tutors. In brief, individuals have different values and priorities, and this conflict of ideas and perspectives leads to the challenge of risk management in chemical-related academic laboratories.

4.4. Continuous Monitoring

Continuous monitoring and regular audits further enhance real-time risk detection [53]. For unexpected laboratory safety and security incidents, effective monitoring and alarm systems can attract the attention of managers at the onset of an incident and help contain the emergency before it escalates. For premeditated laboratory security incidents, effective monitoring and alarm systems can achieve real-time monitoring and timely warning to detect any unauthorized access as well as reduce the physical security risks associated with the “EASY” attribute. In addition, for any sensitive data, hazardous chemicals (e.g., precursors for the manufacture of drugs) or equipment with a high risk of “DESTRUCTIBLE” and “OCCUPIED”, a robust security system is required, which involves all processes of storage, transfer or use of the target [54]. Such monitoring and alarm systems are prevalent in most universities and scientific research institutions. However, whether they can function smoothly is a question worth pondering. For instance, is there a dedicated practitioner actively monitoring on the other end? Are these systems well maintained and regularly inspected? If the security equipment is isolated in the laboratory without attention, it is clear that, even if safety and security incidents occur, there may not be an immediate response.

The development of computer technology and intelligence has provided support for the risk management of chemical-related academic laboratories in colleges and universities. Human–machine collaboration and advanced connectivity are crucial for laboratory safety and security. These intelligent, AI-based machines are reconfigurable, which reduces standardization and complicates risk identification and mitigation [55]. For example, Shu et al. proposed an emergency treatment mechanism for laboratory safety accidents based on the Internet of Things (IoT) and context-aware computing to achieve real-time monitoring of chemistry laboratory environments, timely detection of potential risks and behaviors to reduce laboratory near misses, incidents and accidents [56]. However, the algorithmic bias in computer technology/artificial intelligence deserves attention. For example, machine learning has made rapid progress in recent years and is beginning to have an impact on many disciplines, including chemistry. Deep learning has been successfully applied to organic synthesis analysis [57]. Machine learning algorithms can predict potential risks in scientific research experiments by analyzing historical data (e.g., experimental parameters, expected results, etc.), helping researchers identify risks in advance and take preventive measures. Machine learning can also play a huge role in real-time monitoring of laboratory environments and experimental processes. Artificial intelligence algorithms also provide opportunities for practical chemical experiments. Inherent biases in Artificial Intelligence algorithms, the ambiguity of algorithms and biases in the data used for training can all lead to risk decision errors, including omission errors and execution errors [58].

5. Limitations and Future Directions

This study has several limitations. Firstly, it identifies potential biases in risk research and management within chemical-related academic laboratories and provides insights for future endeavors aimed at narrowing the gap in risk research and management. However, as not all biases are identifiable and solvable, other types of biases not discussed in this paper may affect the integrity of risk research and management and thus deserve further discussion. Secondly, the best practices for reducing the risk biases are worth further exploration, which will provide a better choice for risk management in chemical-related academic laboratories. Thirdly, in the process of safety and security integration, the lack and imbalance of security risks in risk management should be considered. Future studies can further apply the attractiveness assessment and vulnerability assessment of chemical-related academic laboratories, assign scores to laboratory security risk factors in view of international cases of laboratories being attacked by information and physical attacks and then adopt security risk classification management similar to laboratory safety classification management. Fourthly, emerging technologies in the era of digitalization and artificial intelligence can contribute to risk education in chemistry-related academic laboratories, and future studies can further explore and develop in this field. For instance, virtual laboratory teaching based on Web and VR technologies can foster a positive laboratory climate and enhance graduate students’ safety and security awareness. The expectation is to enhance the systematic and foresighted nature of risk management in chemical-related academic laboratories, ensuring the safety and security of laboratory assets and the smooth progression of scientific research endeavors.

6. Conclusions

This paper proposes safety and security biases related to risk management in chemical-related academic laboratories, focusing on systematic bias, bias associated with uncertain risk events, cognitive bias, algorithm/model bias and social/interpersonal bias. These biases can emerge during risk identification, risk assessment, risk control and continuous monitoring. This paper intends to spur discussion on identifying the most critical biases affecting safety and security risk management in chemical-related academic laboratories and on developing more effective methods to recognize and mitigate these biases in risk assessments.

Bias associated with uncertain risk events is the most common and most overlooked type of bias in risk management of chemical-related academic laboratories, mainly involving critical issues/events (particularly the comprehensive safety and security management) that have not been taken into account in risk management. Cognitive bias is mostly related to “human” factors and is one of the most common biases in the risk management of chemical-related academic laboratories. There is a significant correlation between systematic bias and algorithm/model bias in risk research. Fitting with biased data will introduce model bias, which cannot be eliminated and should be reduced as much as possible. Social/interpersonal biases, which mostly involve groupthink perspectives, should not be ignored in the risk management of chemical-related academic laboratories. It is noted that biases are always present and ever-changing, and their understanding may evolve over time. In the future, it is expected to further promote risk community building around defining standards and the best practices for handling biases in safety and security risk management of chemical-related academic laboratories.