Research and Application of Comprehensive Evaluation Model of Campus Safety Culture Based on Game Theory-Regret Theory

Article

Zi-Jing Zhou 1,2 Email15932833691@163.com

Yao Lu 1,2 Email9120100022@jxust.edu.cn

Lin Xie 1,2 Email2120226071@mail.jxust.edu.cn

Hou-dong Liu 1✉ Email7120220071@mail.jxust.edu.cn

Song-Tao Yu 1,2 Emailyusongtao92@163.com

Qian Kang 2✉ Emailkangqianray@126.com

L. X. 1

1 Jiangxi Provincial Key Laboratory of Safe and Efficient Mining of Rare Metal Resource (2023SSY01031) Jiangxi University of Science and Technology 341000 Ganzhou Jiangxi China

2 School of Emergency Management and Safety Engineering Jiangxi University of Science and Technology 341000 Ganzhou China

Zi-Jing Zhou^1,2, Yao Lu^1,2, Lin Xie^1,2, Hou-dong Liu^1,*, Song-Tao Yu^1,^{2 and Qian Kang1,2,*1} Jiangxi Provincial Key Laboratory of Safe and Efficient Mining of Rare Metal Resource (2023SSY01031), Jiangxi University of Science and Technology, Ganzhou Jiangxi 341000, China; 15932833691@163.com(Z.Z.); 9120100022@jxust.edu.cn(Y. L.); 2120226071@mail.jxust.edu.cn (L.X.) ; 7120220071@mail.jxust.edu.cn (H.L.)

² School of Emergency Management and Safety Engineering, Jiangxi University of Science and Technology,

Ganzhou 341000, China; yusongtao92@163.com (S.Y.) ; kangqianray@126.com (Q.K.)

* Correspondence: kangqianray@126.com;7120220071@mail.jxust.edu.cn

Academic Editor(s): Name

Received:

Revised:

Accepted:

Published: date

Copyright: © 2023 by the authors. Submitted for possible open access publication under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

Abstract

To enhance the scientific and objective evaluation of campus safety culture, thereby reducing the risk of campus accidents and improving the overall level of safety culture, this study developed a evaluation index system of campus safety culture from five levels: safety physical culture, safety institutional culture, safety conceptual culture, safety behavioral culture, and safety environmental culture. The campus safety culture evaluation index system employs the ordinal relationship analysis method (G1 method) and the anti-entropy weight method to determine the subjective and objective weights of the indicators, respectively. Additionally, game theory is utilized to compute the comprehensive weights of these indicators. Ultimately, the established comprehensive evaluation model of campus safety culture, founded on game theory and regret theory, was applied to assess the safety culture at universities in Ganzhou. The results show that this evaluation model is suitable for campus safety culture evaluation and provides feasible paths for for such evaluations.

Keywords:

campus safety culture

game theory

regret theory

evaluation

1. Introduction

Safety culture is fundamental to accident prevention and provides essential support for the implementation of safety policies within social governance. The research in this area has significant implications. University campuses, as comprehensive environments that integrate education, scientific research, and daily life, gather students from diverse regions, as well as experts and scholars from various fields, creating a special social environment with dense population and complex personnel. Campus safety has become a critical concern, with frequent incidents drawing significant societal attention [1]. With the expansion of university enrollment, the types and forms of campus safety incidents are continuously evolving, including campus violence, online fraud, traffic safety, and public health events, all of which have garnered widespread concern [2]. As a crucial component of campus culture, campus safety culture plays a vital role in safeguarding the lives and health of teachers and students, protecting property, and maintaining campus stability and harmony. Research on campus safety culture can illuminate the cultural factors underlying these incidents and assist administrators in identifying potential problems. Consequently, developing a scientific and rational evaluation index system for campus safety culture is of great significance for enhancing campus safety management and for the prevention and reduction of various safety incidents.

The study of safety culture has attracted the attention of scholars both domestically and internationally long ago. Utilizing the Analytic Hierarchy Process (AHP) and Fuzzy Comprehensive Evaluation (FCE) theory, Hongxia Li et al. developed an AHP-FCE model to assess the performance of 'zero injury' safety culture construction in coal mines [3]. Similarly, Zhongqing Xie et al. investigated the development of road traffic safety culture [4]. Through expert interviews, they established an evaluation index system for road traffic safety culture construction and the structural validity of the evaluation system was proven to be good based on the Structural Equation Modeling (SEM). Chen Kun et al. introduced an improved Fuzzy Comprehensive Evaluation model based on the SMART safety culture principle, which is employed to assess the safety culture level of petroleum enterprises. The model covers multiple dimensions, including safety materials, behaviors, and systems, and its effectiveness has been validated through case analysis [5]. Additionally, Zhang Su et al. developed a '2–4' model based on the causes of accidents and established a comprehensive index system for evaluating the construction level of laboratory safety culture in colleges and universities. This model was further refined by developing a comprehensive evaluation model based on AHP-FCE to assess the level of laboratory safety culture construction in these institutions [6]. Hu et al. further developed a comprehensive framework for assessing university safety culture, grounded in the "2–4" accident causation model. This framework comprises 4 primary dimensions and 28 sub-dimensions [7]. B. Hesgrove et al. examined the relationship between workplace safety culture and patient safety culture in healthcare institutions by analyzing data from 6,684 respondents [8]. Hammond D. M. addressed this knowledge gap by evaluating the relationship between the safety culture characteristics of U.S. nuclear waste cleanup contractors and two types of performance metrics: personal safety and operational metrics [9]. Orikpete also conducted an analysis of the safety culture in nuclear power plants [10]. Jiao Xiaoyou constructed an intelligent power safety culture evaluation model based on rough set-neural network. The model employs rough set theory to reduce input variables and refine learning samples, followed by the utilization of a neural network to evaluate the power safety culture [11]. Pei Jing-jing and others designed an index system from the perspective of safety culture. They introduced the concept of maturity into the evaluation of safety culture and applied the grey fuzzy comprehensive evaluation method [12]. Gao Shu-xian et al. focuses on the measurement of the safety culture among undergraduate students in a private university in China. By thoroughly exploring the current situation of the students' safety culture in this university, the study analyzes the impacts of various factors on the safety behaviors and safety awareness of college students [13]. Through the comparison of aforementioned research, it reveals that the current focus of safety culture evaluations primarily targets industries such as enterprises, coal mines, nuclear power plants, healthcare, and electric power, with relatively few studies addressing the evaluation of campus safety culture.

The evaluation index of campus safety culture presents a certain degree of ambiguity, making it challenging to express with precise values. Furthermore, the expert decision-making process often exhibits characteristics that lead to avoidance and distortion of judgment. Such subjective decision-making processes may lead to discrepancies between evaluation outcomes and actual conditions. Therefore, when selecting evaluation methods, it is crucial to choose approaches that can transform subjective assessments into objective evaluations, thereby enhancing the objectivity of the results. Regret theory, as a method assuming bounded rationality in decision-makers' behavior, utilizes the ideal evaluation value as a benchmark for distinguishing between subjective and objective assessments. In this context, both the evaluation value and the ideal value undergo utility processing, with the difference between the subjective and objective evaluations being articulated through this disparity. This approach facilitates the conversion of subjective evaluations into objective assessments by correcting the identified differences [14]. Notably, this method offers advantages such as straightforward calculations, strong operability, and results that more accurately reflect the perspectives of decision-makers. In terms of weight calculation, the primary methods for assigning weights to campus safety culture evaluation indicators are the Entropy Method and the Analytic Hierarchy Process (AHP) [15, 16]. However, both the Entropy Weight Method and the Network Analytic Process encounter challenges in effectively obtaining the weight values of the indices, as their calculations may be either subjective or objective.

Compared with traditional research methods, combining the G1 method with the anti-entropy weight method and using game theory to determine the comprehensive weight can more effectively balance the influences of subjective judgment and objective data. Traditional weight determination methods, such as simple subjective weighting methods, overly rely on experts' subjective judgment. They are easily affected by factors like experts' knowledge structures, experiences, and personal preferences, resulting in significant weight deviations. On the other hand, although simple objective weighting methods determine weights based on the degree of variation of the data itself, when the index data fluctuates abnormally, extreme weights will occur, failing to accurately reflect the actual importance of the indexes. The G1 method ranks the importance of indexes based on experts' experience. Its calculation process is simple and intuitive, and it doesn't require complex consistency tests, thus giving full play to experts' advantages in subjective judgment. The anti-entropy weight method overcomes the problem of extreme weights caused by the excessive sensitivity of the degree of disorder of indexes in the entropy weight method, and it can more objectively reflect the information contained in the indexes. By combining these two methods and then optimizing the combination of subjective and objective weights through game theory, the weight determination becomes more scientific and reasonable, and it better meets the actual needs of campus safety culture evaluation. The introduction of regret theory in the construction of the evaluation model fully considers the psychological factors of decision-makers. In the case where campus safety culture evaluation is mostly qualitative, regret theory can effectively avoid the regret psychology of experts during the decision-making process, making the evaluation results closer to the ideal plan. This significantly improves the objectivity and accuracy of the evaluation, providing a brand-new perspective and method for campus safety culture evaluation.

In light of the above, this paper employs game theory and regret theory to evaluate the campus safety culture. It utilizes the Ordinal Relation Analysis Method (G1 method) to calculate the subjective weights of indicators, the anti-entropy weight method to determine the objective weights of indicators, and game theory to compute the comprehensive weights of indicators. Additionally, regret theory is utilized to address the impact of expert regret psychology on decision-making outcomes, thereby enhancing the quality of the evaluation results. Taking three universities in Ganzhou, Jiangxi Province as the research subjects, this study conducts a comprehensive evaluation of campus safety culture, assesses the current state of safety culture construction in these institutions, identifies existing problems and shortcomings, and proposes improvement measures to provide a scientific basis for campus safety management.

2.Construction of campus safety culture evaluation index system

It is crucial to establish a comprehensive evaluation system that objectively reflects the current state of campus safety culture development. The results of this evaluation are closely linked to the selected indicators. Consequently, constructing a set of reasonable evaluation indicators for campus safety culture is crucial. Currently, most scholars in our country categorize campus safety culture into four levels: safety conceptual culture, safety institutional culture, safety behavioral culture, and safety material culture [17]. Furthermore, some researchers classify it into safety material culture, safety institutional culture, safety environmental culture, and safety spiritual culture [15]. Some other scholars have preliminarily constructed the evaluation indexes of campus safety awareness based on the three elements of human, objects, and the environment [18]. Additionally, certain scholars have categorized it into safety concept culture, safety behavior culture, safety management culture, and safety physical education culture [19]. Through a review of multiple sources pertaining to campus safety culture, including literature [3–17], this study identifies the varying research focuses concerning the evaluation of campus safety culture. Building on a thorough examination of relevant literature, as well as conducting field surveys and gathering extensive expert opinions, this article constructs a framework for campus safety encompassing five dimensions: "safety physical culture, safety institutional culture, safety conceptual culture, safety behavioral culture, and safety environmental culture." The comprehensive cultural evaluation index system is illustrated in Fig. 1.

The selected evaluation criteria demonstrate innovation, applicability, and comprehensiveness. Innovatively, this study introduces safety material culture as a distinct dimension, advancing beyond conventional oversimplified categorizations of material factors in prior research. Unlike traditional approaches that broadly group material-related elements, this refined framework captures the diversity and complexity of campus safety infrastructure through sub-indicators such as hardware systems, emergency resources, and engineering technologies. For instance, granular assessment of hardware facilities (e.g., firefighting equipment, surveillance systems) clarifies their operational status, enabling targeted improvements in safety infrastructure.

The framework’s applicability stems from its alignment with the unique context of universities in Ganzhou. These institutions face challenges such as high population density, diverse demographics, and heterogeneous surrounding environments. Through field investigations and consultations with local stakeholders—including faculty, students, safety officers, and experts—the criteria were tailored to address region-specific risks. For example, metrics evaluating transportation safety and neighborhood security management were incorporated to account for the area’s commercial activity and complex public safety conditions, ensuring the assessment reflects on-the-ground realities.

The framework’s comprehensiveness is achieved through a multi-layered structure of 21 indicators across five dimensions. These include: safety material culture (hardware infrastructure), safety institutional culture (management and educational protocols), safety perceptual culture (community awareness and values), safety behavioral culture (normative practices), and safety environmental culture (internal/external environmental factors). These interconnected dimensions collectively address all facets of campus safety culture, mitigating evaluation biases. By holistically analyzing these 21 indicators, the framework systematically evaluates the overall state of campus safety culture, providing actionable insights to identify gaps and prioritize interventions.

Fig. 1

Comprehensive Evaluation Indicators for Campus Safety Culture.

A ₁:Hardware device; A₂:Relief materials; A₃:Engineering technology; A₄:Security system; A₅:Traffic system; B₁:Safety awareness trainning; B₂:Security values; B₃:Propagation of safety concept; B₄:Safety attitude; C₁:Safety management responsibility system; C₂:Safety education system; C₃:Safety inspection system; C₄:Emergency plan system; C₅:Information file management system; D₁:Safety behavior standards; D₂:Safety education activity; D₃:Emergency exercise; D₄:Safety supervision and management; E₁:Food hygiene; E₂:Environmental sanitation; E₃:Hazardous substances management.

2.1Criteria for Indicator Selection

A ₁: It is the material foundation directly related to campus safety. Whether the fire protection facilities, electrical equipment, and sports equipment within the campus are fully equipped and functioning well directly affects the life safety of teachers and students as well as normal campus activities. For example, if fire protection equipment cannot function properly, it will be ineffective in case of a fire, potentially leading to serious consequences; A₂: Adequate and effective rescue materials are the key to dealing with campus emergencies. The reserve of emergency medicines, protective equipment, emergency lighting equipment and other materials determines whether rescue work can be carried out in time at the early stage of an accident to reduce casualties and losses; A₃: The structural safety of campus buildings, the technical reliability of various infrastructure and other engineering technology factors are very important to campus safety. For example, the seismic design of buildings, the reasonable laying of electrical lines and other aspects, if not in line with safety standards, may cause safety accidents; A₄: It includes security systems such as surveillance cameras, access control systems and alarm devices. It is an important means to prevent and monitor campus safety incidents. Surveillance cameras can monitor the activities of people in the campus and abnormal situations in real time, and access control systems can restrict irrelevant personnel from entering the campus and improve campus security; A₅: The road planning, traffic sign setting and transportation management in the campus affect the travel safety of teachers and students. Reasonable road planning and clear traffic signs can help reduce the occurrence of traffic accidents; B₁:A strong safety awareness among teachers and students is the foundation for preventing accidents. By cultivating their understanding and prevention of various safety risks through safety education courses and promotional activities, it can effectively reduce the likelihood of accidents. For example, if students lack safety awareness, they might violate regulations during laboratory classes, leading to danger; B₂: Actively spread the correct safety concept, which can form a good safety culture atmosphere in the campus. For example, advocate the concept of "safety first" and "prevention first", so that the safety awareness is deeply rooted in people's hearts, and teachers and students consciously abide by the safety regulations in their daily behavior; B₃: Safety values reflect the importance and code of conduct of teachers and students on safety. With correct safety values, teachers and students will take the initiative to pay attention to safety issues and actively participate in campus safety management and maintenance work; B₄: The attitude of teachers and students towards safety directly affects their safety behavior. A positive safety attitude is manifested in the conscious compliance with safety regulations and the active investigation of potential safety risks, which helps to create a safe campus environment; C₁: Clarifies the duties of various departments and personnel in campus safety management, serving as an institutional guarantee for the effective implementation of campus safety work. A well-defined responsibility system can prevent gaps in safety management and buck-passing, thereby improving the efficiency and quality of safety management; C₂: Establishes a sound safety education system that specifies the content, methods, frequency, and other aspects of safety education. This ensures the standardization and normalization of safety education efforts, continuously enhancing the safety knowledge and skills of teachers and students; C₃: Conducts regular safety inspections to promptly identify and eliminate potential hazards. Defines the scope, standards, and procedures of inspections to ensure comprehensiveness and effectiveness in safety inspection work, such as periodic checks on teaching buildings, dormitories, cafeterias, and other facilities; C₄: Develops comprehensive emergency response plans, including measures and procedures for handling various emergencies such as fires, earthquakes, and public health incidents. These plans guide teachers and students in conducting rapid and orderly emergency responses during incidents to minimize losses; C₅: Effectively manages campus safety-related information and records, such as safety inspection logs, incident reports, and safety education archives of teachers and students. This information aids in analyzing campus safety conditions, summarizing lessons learned, and providing a basis for safety management decision-making; D₁: Establish clear safety behavioral norms to regulate the daily conduct of faculty and students on campus, such as laboratory operating procedures and public space behavioral guidelines. Standardized behavior can effectively prevent safety incidents; D₂: Organize various safety education activities, such as safety drills and safety knowledge competitions, to enhance the safety awareness and practical skills of faculty and students. Through hands-on activities, they can better master safety techniques and improve their ability to respond to emergencies; D₃: Conduct regular emergency drills to simulate real-life safety incident scenarios, testing and improving the emergency response and coordination capabilities of faculty and students. For example, fire drills help familiarize them with evacuation routes and fire extinguishing techniques; D₄: Implement strict management of hazardous substances on campus, including chemicals, flammable, and explosive materials, covering storage, usage, and transportation. Proper hazardous materials management prevents accidents such as leaks and explosions; E₁: The food hygiene conditions in campus cafeterias directly impact the health of faculty and students. This indicator is selected to evaluate whether food procurement, processing, storage, and other processes meet hygiene standards, preventing food poisoning and other food safety incidents; E₂: A clean and well-maintained campus environment helps prevent the spread of diseases and fosters a comfortable learning and living atmosphere. This includes the cleanliness of public areas and proper waste disposal; E₃: Conduct comprehensive supervision and management of campus safety efforts to ensure the effective implementation of all safety measures. This includes monitoring compliance with safety regulations and inspecting the operational status of safety facilities.

The campus safety culture is categorized into five distinct levels: excellent (Level I), good (Level II), average (Level III), poor (Level IV), and extremely poor (Level V). To effectively pursue the goal of quantitatively analyzing the status of indicators presented in the chart, this study numerically processed each indicator. The quantitative scoring standards for each indicator, along with their corresponding grade classifications, are detailed in Table 1 to facilitate systematic evaluation and comparison.

Table 1

Quantitative Values of Each Grade
Grade	Level Descriptors	Grading Value
Level I	excellent	[8,10]
Level II	good	[6,8)
Level III	average	[4,6)
Level IV	poor	[2,4)
Level V	extremely poor	[0,2)

3.The construction of campus safety culture evaluation model based on game theory-regret theory

3.1 Game theory calculates index weight

3.1.1Calculating the subjective weights of indicators by G1 method

The G1 method is a subjective weighting method proposed by Professor Guo Yajun in 1992, which was built upon the basis of the Analytic Hierarchy Process (AHP). Compared to AHP, the G1 method offers advantages such as straightforward and intuitive calculations, as well as the elimination of the need for consistency testing [20]. The steps of G1 method are as follows:

(1) Determining the order relationship of rating indicators

Let the evaluation index set as

$\left\{ {{X_1},{X_2},{X_3},...,{X_n}} \right\}$

, and invite the experts to rank the importance degree. Firstly, selecting the most important index

$X_{{_{1}}}^{*}$

from each index, and then selecting the relatively important index

$X_{2}^{*}$

from the remaining indexes. By analogy, the order relationship is determined as follows :

$X_{1}^{*} \succ X_{2}^{*} \succ ... \succ X_{n}^{*}$

(2) Determining the relative importance ratio of each index

Inviting experts to judge the importance ratio of two adjacent evaluation indexes

$X_{{_{{k - 1}}}}^{*}$

and

$X_{{_{k}}}^{*}$

, which is recorded as

${r_k}$

${r_k}=\frac{{{w_{k - 1}}}}{{{w_k}}}$

Here, the assignment situation of

${r_k}$

is shown in Table 2 :

Table 2

Assignment situation of
${r_k}$
${r_k}$	Description of valuation
1.0	Indicator $X_{{_{{k - 1}}}}^{}$ is equally important as Indicator $X_{{_{k}}}^{}$ .
1.2	Indicator $X_{{_{{k - 1}}}}^{}$ is slightly more important than Indicator $X_{{_{k}}}^{}$ .
1.4	Indicator $X_{{_{{k - 1}}}}^{}$ is significantly more important than Indicator $X_{{_{k}}}^{}$ .
1.6	Indicator $X_{{_{{k - 1}}}}^{}$ is strongly more important than Indicator $X_{{_{k}}}^{}$ .
1.8	Indicator $X_{{_{{k - 1}}}}^{}$ is extremely more important than Indicator $X_{{_{k}}}^{}$ .
1.1,1.3,1.5, 1.7	Between the two

(3)Determining the weight coefficient of the evaluation index, and according to the above steps, the subjective weight of each evaluation index can be obtained. The formula is as follows :

${w_k}=\frac{1}{{1+\sum\limits_{{k=2}}^{n} {\prod\limits_{{i=k}}^{n} {{r_i}} } }}$			(3)
${w_{k - 1}}={r_k}{w_k}$	(4)

3.1.2 Calculating the objective weight of the index by the anti-entropy weight method

Entropy is the metric of the characterization degree of disorder in a system. The principle of entropy weight method: the greater the degree of the disorder of an index, the smaller the entropy value; consequently, the greater the impact on the evaluation results of the evaluation index system, leading to a higher weight. However, the entropy weight method is susceptible to extreme weights due to the excessive sensitivity of the index's degree of disorder during the weighting process [21].

The anti-entropy weight method is proposed to address the limitations of the entropy weight method. By modifying the entropy weight method’s formula, this improved approach systematically integrates data distribution characteristics and overall context across key steps such as data normalization, indicator proportion calculation, information entropy computation, and weight determination. This reduces excessive sensitivity to disorder levels in weight allocation, thereby yielding more rational and accurate weights that reflect each indicator’s true significance within the evaluation framework. For instance, when applied to campus safety culture assessments, the anti-entropy weight method balances indicator weights effectively, preventing distortions caused by abnormal fluctuations in individual metrics.

To mitigate this issue, this paper employs the anti-entropy weight method to calculate the objective weight of the index. The steps of the anti-entropy weight method are as follows[22]:

(1)data normalization

Normalizing the data for each indicator to prevent the units from affecting the data processing.

For positive indicators :

${u_{ij}}=\frac{{{x_{ij}} - \hbox{min} \left\{ {{x_{1j}},{x_{2j}},......,{x_{nj}}} \right\}}}{{\hbox{max} \left\{ {{x_{1j}},{x_{2j}},......,{x_{nj}}} \right\} - \hbox{min} \left\{ {{x_{1j}},{x_{2j}},......,{x_{nj}}} \right\}}}$	(5)
For negative indicators :

${u_{ij}}=\frac{{\hbox{max} \left\{ {{x_{1j}},{x_{2j}},......,{x_{nj}}} \right\} - {x_{ij}}}}{{\hbox{max} \left\{ {{x_{1j}},{x_{2j}},......,{x_{nj}}} \right\} - \hbox{min} \left\{ {{x_{1j}},{x_{2j}},......,{x_{nj}}} \right\}}}$

(6)

In the formula:

${u_{ij}}$

denotes the normalized data of the

term under the

scheme.

(2)Calculating the proportion of indicators

${p_{ij}}=\frac{{{u_{ij}}}}{{\sum\limits_{{i=1}}^{n} {{u_{ij}}} }}$

(7)

In the formula :

$\:{p}_{ij}$

denotes the index proportion of the

scheme under the

index.(

=1,2, ……, n), ༈

=1,2, ……, m).

(3)Calculating information entropy

${e_j}= - k\sum\limits_{{i=1}}^{n} {{p_{ij}}\ln ({p_{ij}})}$

(8)

Where

$k=\frac{1}{{\ln (n)}}$

, satisfies

${e_j}>0$

In the formula:

${e_j}$

denotes the entropy of the j index.

(4)Calculating Index Weights

${w_j}=\frac{{{e_j}}}{{\sum\limits_{{j=1}}^{m} {{e_j}} }}$	(9)
The calculation results need to satisfy $\sum\limits_{{j=1}}^{m} {{w_j}} =1$
In the formula, ${w_j}$ denotes the weight of index j.( =1,2, ……, )

3.1.3 Calculating the comprehensive weight of indicators by the Game theory

Game theory is a method of operational research used to address conflict situations. Its fundamental concept is to identify an optimal combination of strategies among multiple decision-makers, minimize the discrepancies between their decisions, and maximize overall benefits [23].

The application of game theory in cultural evaluations is methodologically justified. Game theory effectively dissects behavioral interactions and strategic decision-making mechanisms among stakeholders in safety management systems [24]. Campus safety culture assessments involve multiple interdependent indicators with varying significance. Conventional single-weighting approaches fail to account for these complex interdependencies. Game theory treats subjective weights (derived from the G1 method) and objective weights (calculated via the anti-entropy method) as distinct strategic agents. Subjective weights encapsulate expert judgments based on experience and domain knowledge, while objective weights reflect intrinsic data-driven insights. This framework enables the identification of an optimal equilibrium between these dual weighting sources, harmonizing human expertise and empirical evidence. For instance, when weighting "safety institutional culture" and "safety behavioral culture," game theory integrates experts' prioritization of these dimensions with empirical patterns observed in data. This prevents biases from overreliance on subjective judgments or oversights due to purely data-driven approaches, ensuring final weights are both scientifically robust and operationally relevant to campus safety evaluations.

(1)Constructing combined weight vector :

$w_{3}^{T}={A_1}w_{1}^{T}+{A_2}w_{2}^{T}$

(10)

In the formula :

$w_{1}^{T}$

is the subjective weight vector,

$w_{2}^{T}$

is the objective weight vector, and

$w_{3}^{T}$

is the optimal combination weight.

(2)Introducing Game Theory Concepts

$\hbox{min} {\left\| {\sum\limits_{{v=1}}^{2} {{A_v}w_{v}^{T} - w_{z}^{T}} } \right\|_2}\left( {z=1,2} \right)$

(11)

Derivating the above formula, can get the following formula.

$\left[ {\begin{array}{*{20}{c}} {{w_1} \cdot w_{1}^{T}}&{{w_1} \cdot w_{2}^{T}} \\ {{w_2} \cdot w_{1}^{T}}&{{w_2} \cdot w_{2}^{T}} \end{array}} \right] \times \left[ {\begin{array}{*{20}{c}} {{A_1}} \\ {{A_2}} \end{array}} \right]=\left[ {\begin{array}{*{20}{c}} {{w_1} \cdot w_{1}^{T}} \\ {{w_2} \cdot w_{2}^{T}} \end{array}} \right]$

(12)

Solving the preceding equation, we obtain the final combined coefficient

${A_h}$

$\left( {h=1,2} \right)$

(3)Normalizing the coefficients

$A_{h}^{*}=\frac{{\left\| {{A_h}} \right\|}}{{\sum\limits_{{h=1}}^{2} {\left\| {{A_h}} \right\|} }}$	(13)
(4)Getting the optimal combination weight

$w_{3}^{T}=A_{1}^{*}w_{1}^{T}+A_{2}^{*}w_{2}^{T}$

(14)

3.2 regret theory

The core idea of regret theory is to consider the psychological factors of decision-makers, and comparing the selected scheme with alternative options. If the former is superior to the latter, the decision-maker will feel satisfied; Conversely, decision-makers are prone to regret [25]. To a certain extent, this theory enables decision-makers to avoid regret and align their choices more closely with the ideal solution [26]. Because the evaluation of campus safety culture is mostly a qualitative evaluation, the evaluation results are significantly influenced by the behaviors of decision-makers. In contrast, the regret theory model has fewer parameters and are more computationally straightforward, thus providing a more accurate reflection of decision-makers' decision-making processes.

The application of regret theory in campus safety culture evaluation is methodologically justified. During assessments, expert judgments are often influenced by subjective factors such as personal experience and cognitive biases, which may lead to deviations from objective realities. Regret theory addresses this limitation by incorporating psychological dynamics of decision-makers through constructing an ideal reference matrix to compare observed indicator scores against optimized benchmarks. For instance, when evaluating a university's safety culture, experts might disproportionately prioritize training participation rates during scoring. Regret theory quantifies the gap between such subjective weighting and ideal allocations, thereby correcting decision-making biases and aligning results more closely with the actual safety culture status. Furthermore, its computational efficiency allows rapid calibration of multi-indicator, multi-scenario evaluations, significantly enhancing the accuracy and reliability of outcomes. Therefore, applying regret theory in the assessment of campus safety culture can effectively mitigate risks and enhance the objectivity and accuracy of decision-making.

The calculation process for evaluating campus safety culture based on regret theory is as follows [27]:

(1)Constructing evaluation matrix

It is assumed that there are m schemes and n evaluation indexes, selecting n index scores from the m schemes to construct the evaluation matrix N :

$\:\text{P}=\left({\text{p}}_{\text{i}\text{j}}\right)=\left(\begin{array}{ccc}{\text{p}}_{11}&\:\cdots\:&\:{\text{p}}_{1\text{n}}\\\:⋮&\:\ddots\:&\:⋮\\\:{\text{p}}_{\text{m}1}&\:\cdots\:&\:{\text{p}}_{\text{m}\text{n}}\end{array}\right)$

In the formula :

is the score of index n in the scheme m.

(2)Constructing the index ideal matrix I

Let the matrix I represent the ideal matrix for the evaluation objectives. In order to minimize the degree of regret of the decision maker during the decision-making process, the minimum value in each scheme is taken as the ideal value, and the ideal matrix is obtained. From

$\:f\left({x}^{\ast\:}\right)=m$

$\:f\left({x}_{j}\right)$

can see :

$\:{I}_{i}=min\left\{{P}_{ij}\right\}=\left\{{P}_{i}^{\ast\:}\right\}$	(16)
(3)Constructing the index utility value matrix F
The power function and the exponential function are widely utilized as utility functions in economic theory. Currently, the power function, when employed as a utility function, satisfies the hypothesized requirements of regret theory. The formula for the utility function is as follows:

$\:\text{F}={\left({\text{p}}_{\text{i}\text{j}}\right)}^{{\upalpha\:}}$

In the formula :

is the utility function parameter,

(4)Constructing regret-happiness function matrix R

$\:\text{R}=\left[1-{\text{e}}^{-{\upbeta\:}\left({\text{P}}_{\text{i}}^{\ast\:}-{\text{P}}_{\text{i}\text{j}}\right)}\right]$

In the formula :

is a parameter of the regret-joy function,

(5)Constructing the perceived utility value matrix G

$\:\text{G}=\left({\text{g}}_{\text{i}\text{j}}\right)=$

$\:+\text{R}=\left[{\left({\text{P}}_{\text{i}\text{j}}\right)}^{{\upalpha\:}}+\left(1-{\text{e}}^{-{\upbeta\:}\left({\text{P}}_{\text{i}}^{\ast\:}-{\text{P}}_{\text{i}\text{j}}\right)}\right)\right]$

(19)

(6)Calculating the vulnerability assessment value of the criterion layer index

$\:{S}_{t}=k\sum\:_{j=1}^{m}\sum\:_{i=1}^{n}{w}_{ti}^{\ast\:}{g}_{ij}$

In the formula :

is the relative weight of the t criterion index i.

(7)Calculate the vulnerability evaluation value of the evaluation target

$\:\text{S}=\sum\:_{\text{t}=1}^{\text{n}}{\text{w}}_{\text{t}}^{\ast\:}{\text{S}}_{\text{t}}$

In the formula :

represents the relative weight of the t-th criterion indicator i.

4. Using examples

This study selects three universities (K₁, K₂, and K₃) in Ganzhou City, Jiangxi Province as the research objects. Relevant calculations are carried out to determine the safety culture grades of each university and verify the effectiveness and reliability of the model. These three universities cover three typical campus environments, namely the city center, the suburbs, and the urban-rural fringe, aiming to explore the impacts of different geographical locations and public security conditions on safety culture.

The sample selection adheres to the principle of diversity, including a comprehensive university (K₁, with a faculty and student population of approximately 50,000), a science and engineering institution (K₂, which has dense laboratories and has experienced safety accidents in recent years), and a liberal arts college (K₃, featuring a highly open campus and complex surrounding public security).

Data collection is achieved through two channels. The expert scoring is completed by an evaluation team composed of five senior campus safety management experts, including the head of the university safety department, fire protection engineers, emergency management experts, and community safety advisors. The scoring criteria are clarified through discussions. For example, under the dimension of "institutional safety culture", "the emergency response plan is updated at a frequency of ≥ 1 time per year" is considered a full score. For the faculty and student survey, 520 teachers and students are randomly selected for a questionnaire survey. After eliminating invalid data, 478 valid samples are retained. The difference in the scores of core indicators (such as the frequency of emergency drills) between teachers and students and experts is less than 10%, which proves the objectivity of the data. In order to obtain more objective and authentic evaluation scores, this study invited five experts engaged in campus safety management to grade the indicators. These five experts consist of one university professor, one fire protection engineer, one emergency management expert, one community representative, and one safety management expert. Taking into account the actual situations of the three universities in Ganzhou, based on the importance of the indicators, the indicators are graded on a scale of 1 to 10, and the scores of the indicators are directly proportional to their importance levels. The grading results are shown in Table 3.

To ensure the reliability of the experts' scores, this study uses the Spearman correlation coefficient to test the consistency of the scores among the experts. The results show that the five experts have a high degree of consensus on the key indicators: the correlation coefficient among the experts for the institutional safety culture C (such as "Emergency Response Plan System C1") ranges from 0.78 to 0.85 (p < 0.01); the correlation coefficient for the behavioral safety culture D (such as "Emergency Drills D2") is in the range of 0.72 to 0.81 (p < 0.01); the correlation coefficient for the environmental safety culture E (such as "Environmental Hygiene E2") is slightly lower (0.65–0.73, p < 0.05), mainly because environmental indicators are greatly affected by regional differences. The overall average correlation coefficient of all indicators is 0.76 (p < 0.01), indicating that there is a significant consistency in the experts' scores, which supports the credibility of the data.

Table 3

expert scoring results of each index
Universities	Scoring of each index
Serial Number	A₁	A₂	A₃	A₄	A₅	B₁	B₂	B₃	B₄	C₁	C₂	C₃	C₄	C₅
K1	9.6	8.2	8.1	9.5	9.2	8.9	9.3	9.4	8.2	9.5	9.1	9.2	8.7	8.8
K2	9.5	5.2	6.3	7.2	7.8	7.4	6.2	7.1	8.4	7.7	6.3	7.4	7.2	7.3
K3	8.1	6.2	6.5	6.4	6.1	6.6	5.1	6.1	6.2	6.7	5.1	6.3	7.8	8.5
Universities	Scoring of each index
Serial Number	D₁	D₂	D₃	D₄	E₁	E₂	E₃
K1	9.5	9.2	8.5	8.7	8.7	7.7	8.9
K2	8.2	8.3	7.6	8.9	7.9	5.9	7.9
K3	5.1	6.3	7.5	5.8	5.7	6.9	7.8

4.1. Calculating Index Weights

According to the above, this paper uses G1 method, anti-entropy weight method and game theory to calculate the subjective weight, objective weight and comprehensive weight of the index respectively. According to the weighting principle of the index, the specific calculation process is as follows:

4.1.1 Calculating the subjective weights of indicators by G1 method

Due to the excessive number of secondary indicators, it is difficult to assess the significance of each secondary indicator simultaneously. Consequently, a method of calculating weights individually is employed. Ultimately, the weight of the primary indicator is multiplied by the weight of the corresponding secondary indicator to derive the final weight. According to the established index system, the aforementioned scaling method is utilized, and an expert consultation questionnaire survey is conducted. Experts within the relevant field are selected to assign scores to the importance of the indicators. Following this, the scoring results are discussed and synthesized internally, leading to the establishment of a ranking for each index. The relative importance of the indices, post-ranking, is then compared. The subjective weights of the indices, as determined by the G1 method in this study, are presented in Table 4.

Table 4

The final weight determined by
$\:{\text{G}}_{1}$
method
first grade indexes	weight	second index	local weight	global weight
A	0.1574	A₁	0.2340	0.0368
		A₂	0.2574	0.0405
		A₃	0.1950	0.0307
		A₄	0.1773	0.0279
		A₅	0.1364	0.0215
B	0.2250	B₁	0.3271	0.0736
		B₂	0.2726	0.0613
		B₃	0.2097	0.0472
		B₄	0.1906	0.0429
C	0.2046	C₁	0.2677	0.0548
		C₂	0.2028	0.0415
		C₃	0.1560	0.0319
		C₄	0.2434	0.0498
		C₅	0.1300	0.0266
D	0.2700	D₁	0.2387	0.0645
		D₂	0.1836	0.0496
		D₃	0.2626	0.0709
		D₄	0.3151	0.0851
E	0.1430	E₁	0.3860	0.0552
		E₂	0.3216	0.0460
		E₃	0.2924	0.0418

4.1.2 Calculating the objective weight of the index by the anti-entropy weight method

Utilizing formulas (5) and (6) to normalize the data for each indicator across the three universities. Subsequently, calculating the proportion matrix of the normalized data using formula (7). Finally, applying formulas (8) and (9) to compute the proportion of each indicator, as well as to determine the information entropy and weights necessary for establishing the objective weight of each indicator, as illustrated in Table 5.

Table 5

The final weight determined by the anti-entropy weight method
first grade indexes	weight	second index	weight
A	0.2283	A1	0.0572
		A2	0.0469
		A3	0.0278
		A4	0.0425
		A5	0.0539
B	0.1927	B1	0.0476
		B2	0.0427
		B3	0.0453
		B4	0.0571
C	0.2485	C1	0.0480
		C2	0.0451
		C3	0.0489
		C4	0.0497
		C5	0.0568
D	0.1954	D1	0.0561
		D2	0.0559
		D3	0.0262
		D4	0.0572
E	0.1351	E1	0.0563
		E2	0.0540
		E3	0.0248

4.1.3 Calculating the combined weight by the Game theory

Using the formula (10) -formula (14), the optimal coefficients

$A_{1}^{*}=0.5847$

$A_{2}^{*}=0.4153$

are obtained. The combined weights of each index can be obtained, as shown in Table 6 :

Table 6

Combined weight
	objective weights	subjective weights	combination weight
A1	0.0572	0.0368	0.0453
A2	0.0469	0.0405	0.0431
A3	0.0278	0.0307	0.0295
A4	0.0425	0.0279	0.0339
A5	0.0539	0.0215	0.0349
B1	0.0476	0.0736	0.0628
B2	0.0427	0.0613	0.0536
B3	0.0453	0.0472	0.0464
B4	0.0572	0.0429	0.0488
C1	0.0480	0.0548	0.0520
C2	0.0451	0.0415	0.0430
C3	0.0489	0.0319	0.0390
C4	0.0497	0.0498	0.0498
C5	0.0568	0.0266	0.0391
D1	0.0561	0.0645	0.0610
D2	0.0559	0.0496	0.0522
D3	0.0262	0.0709	0.0523
D4	0.0572	0.0851	0.0735
E1	0.0563	0.0552	0.0557
E2	0.0540	0.0460	0.0493
E3	0.0248	0.0418	0.0347

4.2 Calculating the Campus Safety Culture Assessment System

According to the principles of regret theory, the first step involves evaluating the indicator matrix to establish the ideal point matrix. Subsequently, the utility value matrix, regret-pleasure value matrix, and perceived utility matrix are calculated to derive the evaluation value for each criterion layer. Finally, the overall evaluation is determined through the integration of comprehensive weights and evaluation values. Due to space limitations, this article will illustrate the calculation process using 'A' as an example. The specific calculation process is detailed below:

(1)According to the evaluation expert index score in Table 4, the original scoring matrix

$\:{Q}_{A}$

of the ' safety material culture A ' criterion layer is constructed:

${{\varvec{Q}}_A}=\left[ {\begin{array}{*{20}{r}} {9.6000}&{8.2000}&{8.1000}&{9.5000}&{9.2000} \\ {9.5000}&{5.2000}&{6.3000}&{7.2000}&{7.8000} \\ {8.1000}&{6.2000}&{6.5000}&{6.4000}&{6.1000} \end{array}} \right]$

(2)Determining the ideal point matrix

According to the formula (16), the minimum value of each evaluation index is selected as the ideal point matrix

$\:{P}_{A}$

${{\varvec{P}}_A}={\left[ {8.1000\qquad 5.2000\qquad 6.3000\qquad 6.4000\qquad 6.1000} \right]^T}$

(3)Utility matrix

According to Eq. (17), the utility matrix

$\:{H}_{A}$

can be obtained.

${{\varvec{H}}_A}=\left[ {\begin{array}{*{20}{r}} {9.3853}&{8.0293}&{7.9323}&{9.2885}&{8.9981} \\ {9.2885}&{5.1150}&{6.1851}&{7.0593}&{7.6414} \\ {7.9323}&{6.0879}&{6.3795}&{6.2823}&{5.9907} \end{array}} \right]$

(4)Regret - Delight Value Matrix

According to Eq. (18), the regret-happiness matrix

$\:{R}_{A}$

is :

${{\varvec{R}}_A}=\left[ {\begin{array}{*{20}{r}} { - 0.0146}&{ - 0.0296}&{ - 0.0176}&{ - 0.0305}&{ - 0.0305} \\ { - 0.0137}&{0.0000}&{0.0000}&{ - 0.0078}&{ - 0.0166} \\ {0.0000}&{ - 0.0098}&{ - 0.0019}&{0.0000}&{0.0000} \end{array}} \right]$

(5)Perceptual utility matrix

According to Eq. (19), the perceived utility matrix

$\:{D}_{A}$

is :

${{\varvec{D}}_A}=\left[ {\begin{array}{*{20}{r}} {9.3707}&{7.9997}&{7.9147}&{9.2580}&{8.9676} \\ {9.2749}&{5.1150}&{6.1851}&{7.0515}&{7.6248} \\ {7.9323}&{6.0781}&{6.3775}&{6.2823}&{5.9907} \end{array}} \right]$

(6)Calculation of comprehensive risk assessment value

Similarly, the perceived utility matrices of B, C, D and E are :

${{\varvec{D}}_B}=\left[ {\begin{array}{*{20}{r}} {8.6850}&{9.0533}&{9.1592}&{8.0097} \\ {7.2456}&{6.0771}&{6.9524}&{8.2015} \\ {6.4766}&{5.0176}&{5.9907}&{6.0879} \end{array}} \right]$

${{\varvec{D}}_C}=\left[ {\begin{array}{*{20}{r}} {9.2610}&{8.8616}&{8.9696}&{8.4992}&{8.5960} \\ {7.5347}&{6.1734}&{7.2426}&{7.0593}&{7.1563} \\ {6.5738}&{5.0176}&{6.1851}&{7.6356}&{8.3083} \end{array}} \right]$

${{\varvec{D}}_D}=\left[ {\begin{array}{*{20}{r}} {9.2449}&{8.9696}&{8.3103}&{8.4853} \\ {7.9987}&{8.1066}&{7.4464}&{8.6770} \\ {5.0176}&{6.1851}&{7.3504}&{5.6989} \end{array}} \right]$

${{\varvec{D}}_E}=\left[ {\begin{array}{*{20}{r}} {8.4843}&{7.5268}&{8.6968} \\ {7.7168}&{5.7962}&{7.7374} \\ {5.6017}&{6.7582}&{7.6414} \end{array}} \right]$

(7)Evaluation results of safety culture in colleges and universities

According to Eq. (21), the overall scores of the three universities can be obtained, as shown in the following table :

Table 7

College score
University number	score	rank
K1	8.6845	1
K2	7.3258	2
K3	6.3295	3

As shown in Table 1, University

$\:{K}_{1}$

has an excellent overall safety culture evaluation, while Universities

$\:{K}_{2}$

and

$\:{K}_{3}$

have good overall safety culture evaluations. Based on the weight calculations and assessment results, the main factors in the university safety culture assessment include safety awareness training, safety values, safety management responsibility systems, safety behavior norms, safety education activities, emergency drills, safety supervision and management, and food hygiene, among others. Therefore, universities should prioritize the aforementioned indicators and take targeted measures to optimize and improve, thereby enhancing safety levels and reducing casualties or property damage.

Combined with the field investigation of University

$\:{K}_{1}$

, it can be seen that the university has performed well in the construction of safety physical culture, safety conceptual culture, safety institutional culture, safety behavioral culture, and safety environmental culture. The university has established a comprehensive security system and has invested in a significant amount of hardware, ensuring coverage across the entire campus. This includes the installation of advanced monitoring systems and alarm mechanisms, which facilitate the prompt identification and management of security threats. Furthermore, the university prioritizes the cultivation of safety values for all members of the school and emphasizes the dissemination of safety concepts. It has established safety education courses and lectures, and through regular safety training and drills, it enhances the safety knowledge and emergency response capabilities of faculty and students. Additionally, the university has established and refined its safety management accountability system and safety inspection protocols, developed a robust emergency response plan, and defined appropriate safety behavior standards. Regular safety education initiatives are also conducted. Overall, the university's safety culture is excellent, and there have been virtually no safety incidents in the past three years.

According to the analysis of the survey results of University

$\:{K}_{2}$

, the university has been equipped with a considerable number of advanced hardware equipment, showing a good attitude of safety management. The university routinely organizes safety education initiatives and demonstrates effective performance in safety supervision and management. In general, the level of safety culture construction in this university is relatively high. However, it is important to acknowledge that over the past three years, the university has experienced a limited number of safety incidents, including a laboratory safety mishap and an incident of campus violence. Consequently, to enhance the effectiveness of safety management further, the university should focus on establishing and refining its safety education framework, ensuring an adequate supply of emergency response materials, and bolstering the development of campus infrastructure [28]. Additionally, it is essential to promote the cultivation of safety values among students, actively disseminate safety concepts, and enhance the cleanliness and maintenance of the campus environment.

According to the analysis of the survey results of University

$\:{K}_{3}$

, the university is equipped with relatively complete hardware resources. It has established a thorough emergency plan system and an effective information file management system, and regularly carried out emergency drill activities. Furthermore, the university demonstrates effective management of hazardous materials. However, there remain notable deficiencies in other aspects of safety management. Overall, the university's safety culture is good, but there is still a gap compared to the first two universities. In the past three years, the university has experienced a series of safety incidents, including but not limited to a fire accident, a traffic accident, and a case of school bullying. To enhance the level of the safety culture further, it is recommended that the university intensifies its safety education and training initiatives to improve the safety awareness and emergency response capabilities of both faculty and students. Additionally, it is necessary to refine the safety management system, intensify safety inspections and hazard identification efforts, and ensure that all safety measures are effectively implemented. Concurrently, the university should prioritize the promotion and dissemination of safety culture, using various methods and formats to instill safety concepts deeply and create a safety culture atmosphere involving all faculty and students.

4.3 Result analysis

To improve the objectivity and reliability of evaluation outcomes, this study employs a weight distribution methodology that integrates both subjective and objective factors. Specifically, it utilizes the G1 method, the anti-entropy weight method, and game theory to ascertain the subjective, objective, and comprehensive weights of the indices, respectively. For example, in calculating the weights of first-level indicators, if only the entropy weight method is used to compute the objective weights for "Safety Physical Culture A","Safety Conceptual Culture B","Safety Institutional Culture C","Safety Behavioral Culture D" and "Safety Environmental Culture E", the resulting weights are (0.2283, 0.1927, 0.2485, 0.1954, 0.1351), respectively. In this case, the weight of "Safety Environmental Culture E" is significantly underestimated, which could overlook the critical role of the safety environment in campus safety. In practice, building a safe environment is fundamental to maintaining university safety. A good safety environment can prevent potential risks, thereby protecting the life and property safety of teachers and students, and ensuring the normal operation of the campus and the smooth progress of educational activities.

By incorporating the G1 method, experts can use their professional experience and judgment to rank the importance of each indicator, thereby determining subjective weights. Subsequently, using game theory concepts, the comprehensive weights of the first-level indicators are calculated, resulting in weights of (0.1868, 0.2116, 0.2228, 0.2390, 0.1398). The advantage of this method is that it not only considers the objective attributes of the indicators themselves but also makes full use of experts' subjective judgments, making the calculated weights more reflective of the actual situation, thereby enhancing the accuracy and reliability of the assessment. By utilizing this comprehensive weighting approach, the actual significance of each index within the context of campus safety culture can be more effectively represented, thereby facilitating more informed decision-making in campus safety management.

Following the assessment process of regret theory, this model demonstrates its comprehensiveness and depth in evaluating campus safety culture. This model not only offers a macro-level overview of the overall safety culture but also provides an in-depth analysis of each criterion layer. The detailed evaluation outcomes serve as a foundation for identifying and analyzing issues present within each criterion layer, thereby facilitating the development of targeted solutions aimed at enhancing the emergency rescue system at a macro level.

The specific evaluation results indicate that the weighted sum evaluation values for University

$\:{K}_{1}$

, University

$\:{K}_{2}$

, and University

$\:{K}_{3}$

are 8.6845, 7.3258, and 6.3295, respectively, categorizing them as excellent, good, and good. University K₁ demonstrated excellence across all safety culture dimensions, with scores of 8.7078 (Material Safety Culture A), 8.7472 (Conceptual Safety Culture B), 8.8535 (Institutional Safety Protocols C), 8.7196 (Behavioral Safety Practices D), and 8.2384 (Environmental Safety Standards E). The standout performance in Institutional Safety Protocols (C) highlights K₁’s robust framework for policy development, implementation, and oversight. This systematic approach ensures rigorous compliance with safety regulations, effectively mitigates risks, and fosters a harmonized safety culture where all dimensions reinforce one another.

In contrast, University K₂ achieved an overall score of 7.3258, classified as "satisfactory." While scoring relatively higher in Behavioral Safety Practices (D: 8.0799)—reflecting effective emergency drills, safety training, and incident response protocols—critical gaps persist. Weaknesses in Institutional Safety Protocols (C) and Conceptual Safety Culture (B) suggest inconsistent policy enforcement and underdeveloped safety awareness. For instance, incomplete safety education systems may hinder comprehensive risk knowledge among stakeholders, and superficial safety values among some students correlate with preventable incidents. Staff-related vulnerabilities are particularly evident: lapses in laboratory safety supervision (e.g., inadequate adherence to operational guidelines) and insufficient monitoring of student interactions have contributed to equipment mishaps and unresolved campus conflicts [29]. These findings underscore the need for K₂ to strengthen institutional accountability and cultivate deeper safety consciousness across its community.

University K₃ achieved an overall score of 6.3295, classified as "satisfactory," with Institutional Safety Culture (C) identified as its strongest dimension. This reflects foundational capabilities in policy development, such as established emergency response protocols, information management systems, and regular safety drills. However, K₃ experienced recurring safety incidents over the past three years, including fires, traffic accidents, and campus bullying. Further analysis reveals critical gaps in Material Safety Infrastructure (e.g., aging equipment and inadequate security system coverage) and Behavioral Safety Practices (e.g., inconsistent enforcement of traffic/fire safety protocols and insufficient risk awareness among stakeholders). These systemic weaknesses in infrastructure maintenance and behavioral compliance directly correlate with the observed incident patterns, underscoring the urgency of targeted interventions in these areas.

The strong correlation between Institutional Safety Protocols (C) and Behavioral Safety Practices (D) (r = 0.73, p < 0.01) underscores the regulatory role of formal policies in shaping safety behaviors. For instance, University K₁’s mandatory monthly emergency drills (D2) directly correlated with exceptional behavioral compliance scores (D1 = 9.5, D2 = 9.2), demonstrating how systematic enforcement drives measurable improvements. While Environmental Safety Standards (E) carried the lowest weight (0.1398) in the evaluation framework, K₁’s strategic investments in infrastructure—such as achieving 92% smart surveillance coverage—yielded a high E-score of 8.24, suggesting that practical management priorities can dynamically offset theoretical weightings.

Cluster analysis revealed a stark contrast between K₁’s rigorous policy implementation (C = 8.85) and the suboptimal environmental management at K₂/K₃ (E = 6.32), reinforcing institutional rigor as the linchpin of campus safety. The dominance of Behavioral Practices (D) and Institutional Protocols (C) across evaluations highlights their dual role as cornerstones of safety culture. To institutionalize safety, universities must prioritize (1) continuous safety education to strengthen risk awareness and (2) accountability mechanisms to ensure policy adherence.

Notably, while Environmental Safety (E) held lesser weight, its practical impact remains critical. Proactive measures—such as upgrading aging infrastructure, optimizing security systems, and collaborating with local authorities to address off-campus risks—are essential to mitigate incidents like those observed at K₃ (e.g., fires, traffic accidents) [30]. These findings advocate for a balanced approach: leveraging institutional frameworks to standardize behaviors while contextually addressing environmental vulnerabilities.

By comparing the evaluation results of the three universities, it is found that different universities have both common problems and unique characteristics in the construction of safety culture. The common problems include the need to strengthen the cultivation of safety awareness and the improvement of the implementation intensity of safety systems. Regarding these common issues, universities can enhance communication and cooperation, share successful experiences, and jointly improve the level of campus safety culture construction.

As for their respective characteristics, for example, University K₁ has a relatively balanced development in all aspects. University K₂ has an advantage in the aspect of safety behavior culture but has potential safety accident risks. University K₃ has a certain foundation in safety institutional culture, yet problems are prominent in some aspects. Each university should formulate personalized improvement measures according to its own actual situation, and strengthen the construction of weak links in a targeted manner, so as to achieve the continuous optimization of campus safety culture construction.

7. Conclusion

1) In this study, the G1 method is applied to determine the subjective weights of the indicators, and the anti-entropy weight method is used to quantify the objective weights of the indicators. Finally, game theory is integrated to synthesize these weights, effectively balancing the influences of subjective judgment and objective data. This comprehensive weighting strategy prevents the single-sided tendency in the process of weight determination, whether it is over-reliance on subjective opinions or complete dependence on objective data, thus enhancing the credibility and robustness of the evaluation results. Meanwhile, considering the complex and systematic characteristics of the evaluation of the campus safety culture system, and that most of the evaluation indicators are qualitative, a comprehensive evaluation model is established by combining game theory and regret theory. From five aspects, namely "physical safety culture", "conceptual safety culture", "institutional safety culture", "behavioral safety culture", and "environmental safety culture", 21 indicators are selected to construct the evaluation index system of campus safety culture. Through a comparative study with the set pair analysis method and the fuzzy comprehensive evaluation model, it is found that the calculation results of the model established in this paper are reliable and reasonable, which can provide a new idea and method for the evaluation of campus safety culture. At the same time, considering the complexity and systematic nature of campus safety culture system evaluation, and given that most evaluation indicators are qualitative, a comprehensive evaluation model is established by combining game theory and regret theory. From five aspects—"safety material culture," "safety conceptual culture," "safety institutional culture," "safety behavioral culture," and "safety environmental culture"—21 indicators are selected to construct a campus safety culture evaluation index system. Comparative studies with the set pair analysis method and the fuzzy comprehensive evaluation model reveal that the model established in this paper yields reliable and reasonable calculation results, providing a new approach and methodology for campus safety culture evaluation.

2) In this study, three schools were selected for analysis. Compared with the approach of evaluating only a single school in previous studies, the multi-case analysis method can better verify the universality and stability of the model. By comparing the safety culture situations of different schools, it is possible not only to identify the common problems in the construction of safety culture among these schools but also to put forward personalized improvement suggestions tailored to the characteristics of each school. Based on the indicator system established in this study, specific suggestions are proposed to enhance the level of campus safety culture. Universities should regularly maintain and update hardware facilities, rationally plan the transportation system, and ensure an adequate supply of rescue materials. Safety education should be integrated into teaching, and safety concepts should be disseminated through multiple channels to cultivate correct safety values. All kinds of safety systems should be improved, responsibilities should be clearly defined, supervision and assessment should be strengthened, the forms of safety education should be enriched, and the inspection system should be strictly implemented. The safety behaviors of teachers and students should be standardized, with clear rewards and punishments. Educational activities and drills should be carried out regularly, and the management of hazardous substances should be enhanced. The food hygiene on campus should be strictly supervised, environmental hygiene management should be strengthened, the safety supervision mechanism should be improved, and a safe campus environment should be comprehensively created.

3) This study has certain limitations. On one hand, the scope of data collection is relatively limited. Only three universities in Ganzhou, Jiangxi Province, were selected as the research objects. The sample is rather homogeneous in terms of geography and type, which may not fully represent the campus safety culture situations of universities in different regions and of different types. Thus, the universality of the research results needs to be further improved. On the other hand, although various factors were comprehensively considered during the construction of the indicator system, with the continuous changes in the campus safety situation, new safety issues may keep emerging. The existing indicator system may not be able to cover all important factors in a timely manner and requires continuous updating and improvement.

In addition, during the process of model construction in this study, the determination of some parameters relied on experts' experience and subjective judgment. Despite the adoption of certain methods for optimization, there may still be some subjective biases. Future research can be carried out in the following directions: Firstly, expand the sample size and scope, and select universities from different regions, at different levels, and of different types for research to further verify the universality and stability of the model and enhance the representativeness of the research results. Secondly, closely monitor the new developments and issues in the field of campus safety, and update and improve the evaluation indicator system in a timely manner. For example, consider incorporating relevant indicators from emerging fields such as network security and artificial intelligence security into the system to make the evaluation results more reflective of the actual situation of campus safety culture. Thirdly, explore more scientific and objective methods for determining model parameters to reduce the influence of subjective factors and improve the accuracy and reliability of the model. For instance, technologies such as big data analysis and machine learning can be integrated to optimize and adjust the model parameters. Fourthly, conduct in-depth research on the dynamic change laws of campus safety culture, establish a dynamic evaluation model, achieve real-time monitoring and early warning of campus safety culture, and provide more timely and effective decision-making support for campus safety management.

Author Contribution

Author Contributions: Z.Z.: investigation, and editing; Y.L.: data curation; L.X.: writing—original draft; H.L.: software and writing—review and editing; S.Y.: formal analysis and validation; Q.K.: conceptualization and formal analysis. All authors have read and agreed to the published version of the manuscript.

Funding:

This research was supported by Jiangxi Provincial Social Science Youth Foundation (Grants No. 24GL68D), Humanities and Social Sciences Research Project of Jiangxi Province's Universities (Grants No. JC23216).

Institutional Review Board Statement: Not applicable

Informed Consent Statement: Not applicable

Data Availability Statement: Not applicable.

Conflicts 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.

References

GONG Yun-hua. Safety culture among Chinese undergraduates: A survey at a university. Saf. Sci. 111, 17–21 (2019).

Alqahtani, M. A. Factors affecting cybersecurity awareness among university students. Appl. Sci. 12 (5), 2589 (2022).

LI Hong-xia, D. I., Hong-xi & Xu-an, W. A. N. G. An AHP Based Study of Coal-Mine Zero Harm Safety Culture Evaluation. Advances on P2P, Parallel, Grid, Cloud and Internet Computing: Proceedings of the 11th International Conference on P2P, Parallel, Grid, Cloud and Internet Computing (3PGCIC–2016) November 5–7, 2016, Soonchunhyang University, Asan, Korea. Springer International Publishing, (2017).

Zhong-qing, X. I. E. & Pei-lin, Z. H. A. N. G. Evaluative study of the culture construction of road traffic safety based on structural equation modeling. 2019 5th international conference on transportation information and safety (ICTIS). IEEE, (2019).

Kun, C. H. E. N. et al. Safety culture assessment of petroleum enterprises based on SMART. Chin. J. Geochem. 32, 273–280 (2013).

Su, Z. H. A. N. G., Ya-xian, W. A. N. G. & Jin-gui, W. A. N. G. Fuzzy Comprehensive Evaluation of the Construction Level of Laboratory Safety Culture in Colleges and Universities Based on the 2–4. Model. Experimental Technol. Manage. 38, 7 (2021).

Rui-li, H. U., Ye, Z. H. A. N. G. & Bei-fang, G. U. Comprehensive Evaluation of University Safety Culture Construction Level Based on 2–4. Model. Int. J. Environ. Res. public. health. 19 (23), 16145 (2022).

Hesgrove, B. et al. Associations between patient safety culture and workplace safety culture in hospital settings. BMC Health Serv. Res. 24 (1), 568 (2024).

Hammond, D. M. et al. Understanding the relationship between safety culture and safety performance indicators in US nuclear waste cleanup operations. Saf. Sci. 166, 106241 (2023).

Orikpete, O. F. & Ewim, D. R. E. Interplay of human factors and safety culture in nuclear safety for enhanced organisational and individual Performance: A comprehensive review. Nucl. Eng. Des. 416, 112797 (2024).

Xiao-you, J. I. A. O., Zhao-guang, H. U. & Shou-xin, S. O. N. G. Intell. power Saf. Cult. Assess. model. Simul. Anal. [J] J. Beijing Jiaotong Univ. 32.2 : 48–52. (2008).

Pei, J., Liu, L., Chi, Y. & Yu, C. Research on the Maturity Evaluation Model of Enterprise Safety Culture. Int. J. Environ. Res. Public Health. 20 (3), 2664 (2023).

Shu-xian, G. A. O. et al. Safety culture measurement among Chinese undergraduates at a private university: development and validation. Front. public. health. 10, 825106 (2022).

WEN & Zhi LIAO Hu-chang, and José Rui Figueira. A preference disaggregation-driven multiple criteria sorting model based on regret theory. Omega 129, 103150 (2024).

Fu-qiang, Y. A. N. G. & Xiao-xia, L. I. U. ZHU Wei-fang. Entropy weighted matter-element extension model for college campus safety culture evaluation and its application[J]36 (Research & Exploration in Laboratory, 2017). 6.

Fu-qiang, Y. A. N. G. LIN Zi-yi. Application of ANP-SWOT Model in the Construction of Safety Culture on University Campus[J] (Safety and Environmental Engineering, 2020).

JIN Xiao-hua. Research on the Construction of Campus Safety Culture System in Colleges and Universities. Educ. Informatization Forum, pp. 42–44. (2022).

Shang, H., Wang, W. & Zhao, M. Construction and Evaluation of Influencing Factor Evaluation System Model for Campus Safety Awareness. Saf. Manage. Hum. Factors. 64, 64 (2022).

Chen, X. Wu Zhi-qiang, Zhang Jun, Sensitivity Analysis of Campus Safety Culture Based on Fuzzy Analytic Hierarchy Process. In 2020 Chinese Automation Congress (CAC) (2539–2543). IEEE. (2020), November.

Gen, L. I. et al. Fuzzy Comprehensive Evaluation of Pilot Cadets’ Flight Performance Based on G1 Method. Appl. Sci. 13.21 (2023): 12058 .

Jian-jun, Y. A. N. G. et al. Mathematical Problems in Engineering Decision-Making Based on Improved Entropy Weighting Method: An Example of Passenger Comfort in a Smart Cockpit of a Car. Mathematical Problems in Engineering. 1 (2022): 6846696. (2022).

Wan-yu, W. A. N. G. et al. Multi-criteria evaluation of distributed energy system based on order relation-anti-entropy weight method. Energies 14.1 : 246. (2021).

Sohrabi, M., Karim & Azgomi, H. A survey on the combined use of optimization methods and game theory. Arch. Comput. Methods Eng. 27 (1), 59–80 (2020).

Meng, Q. et al. A review of game theory application research in safety management. IEEE access. 8, 107301–107313 (2020).

Coricelli, G. et al. Regret and its avoidance: a neuroimaging study of choice behavior. Nat. Neurosci. 8 (9), 1255–1262 (2005).

Bleichrodt, H. & Peter, P. Wakker. Regret theory: A bold alternative to the alternatives. Econ. J. 125 (583), 493–532 (2015).

Zhou Huan, W. A. N. G., Jian-qiang & Hong-yu, Z. H. A. N. G. Grey stochastic multi-criteria decision-making based on regret theory and TOPSIS. Int. J. Mach. Learn. Cybernet. 8, 651–664 (2017).

Viitaharju, P., Yliniemi, K., Nieminen, M. & Karttunen, A. J. Learning experiences from digital laboratory safety training. Educ. Chem. Eng. 34, 87–93 (2021).

Shusko, R. How Do the Perceptions and Experiences of Faculty and Staff Influence the Development and Shaping of the Safety Culture at a Community College? A Qualitative Case Study (Diss. Frostburg State University, 2024).

Cuyjet, M. J. & Meriwether, J. L. Environmental influences on college culture. In Multiculturalism on Campus (40–65). Routledge. (2023).

Disclaimer/Publisher’s Note. The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Data Availability

Data availability : The datasets used or analyzed in this study are available from the corresponding author on reasonable request.

Yes