Unsafe behavior assessment method based on structural equation model

Abstract

The invention relates to the technical field of unsafe behavior evaluation, and provides an unsafe behavior evaluation method based on a structural equation model, which comprises the following steps: step 1, analyzing influence factors, step 2, proposing hypotheses, step 3, dividing indexes, step 4, performing questionnaire survey, step 5, preprocessing data, step 6, checking reliability and validity, and step 7, evaluating unsafe behaviors. The method comprises the following steps of 1, constructing a structural equation model, 8, judging whether indexes reach the standard or not, 9, outputting a standardized path or correcting the structural model, 10, re-checking the corrected model, and 11, analyzing the path. According to the method, influence factors causing unsafe behaviors of workers are identified and analyzed, influence relations among the influence factors are assumed, then the influence factors are divided into different potential variables, observation variables corresponding to the potential variables are constructed, a questionnaire table is formulated to investigate the workers, a structural equation model is constructed according to data passing trust-validity test, and the work efficiency of the workers is improved. Then performing model fitting degree and saliency inspection, if the requirements are met, obtaining a standardized path coefficient and saliency, if the requirements are not met, correcting the model, then performing model fitting degree and saliency inspection again until the requirements are met, and analyzing the path of the model meeting the requirements; and finally, obtaining positive and negative influence relationships between objective and scientific influence factors and positive and negative influence relationships of the influence factors on the unsafe behaviors, and providing suggestions for inhibiting the unsafe behaviors.

Description

Technical Field

[0001] The present invention relates to the technical field of unsafe behavior assessment, and more specifically, to an unsafe behavior assessment method based on a structural equation model. Background Art

[0002] With the progress of society and people's increasing attention to safety, coal mine safety has risen to the national strategic level. The government, enterprises and all sectors of society are constantly strengthening the supervision of coal mine safety production, promoting the advancement of safety technology, reducing accidents and protecting the rights and interests of miners. However, due to the complex operating environment of coal mines and the large number of operators, it is particularly important to prevent the occurrence of unsafe behaviors of people or unsafe conditions of things.

[0003] Behavioral safety theory believes that unsafe behaviors are the main cause of accidents, and the accident rate can be significantly reduced by identifying and changing these behaviors. Heinrich's accident causation theory believes that the occurrence of accidents is the result of a series of events, including genetic factors and social environment, human shortcomings, unsafe behaviors of people or unsafe conditions of things, accidents themselves, and injuries, and unsafe behaviors of people or unsafe conditions of things are the direct causes of accidents. Although theoretical research on the mechanism of accident occurrence caused by unsafe human behaviors has become more complete and mature, due to the complexity and dynamic nature of unsafe human behaviors and the many factors that affect the occurrence of unsafe behaviors, theoretical research may not be suitable for practice. Therefore, the evaluation model and empirical analysis of unsafe behaviors are crucial, and then appropriate measures can be taken to suppress unsafe behaviors and achieve safe production in enterprises;

[0004] To solve the above problems, this application proposes an unsafe behavior assessment method based on a structural equation model. Summary of the invention

[0005] The purpose of the present invention is to provide an unsafe behavior assessment method based on a structural equation model, which analyzes the positive and negative influence relationships between the influencing factors that lead to the unsafe behavior of workers and the positive and negative influence relationships of the influencing factors on the unsafe behavior through the structural equation model, so as to provide suggestions for suppressing unsafe behavior and solve the problems in the prior art.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A method for evaluating unsafe behavior based on a structural equation model includes the following steps:

[0008] Step 1: Analyze influencing factors, identify and analyze the influencing factors that lead workers to unsafe behaviors;

[0009] Step 2: Propose a hypothesis, a hypothesis about the influence relationship between influencing factors and unsafe behaviors and a hypothesis about the influence relationship between influencing factors;

[0010] Step 3: Divide the indicators. According to the unsafe behavior evaluation index system, divide the influencing factor indicators into different latent variables and construct different observed variables corresponding to different latent variables. Take the latent variables as the primary indicators and the observed variables corresponding to each latent variable as the secondary indicators.

[0011] Step 4: Questionnaire survey. Based on the observed variables corresponding to the different latent variables in step 3, a scale evaluation set of factors affecting unsafe behaviors is designed and compiled, and a questionnaire is constructed. The questionnaire is distributed to employees for investigation. The questionnaire is a Liket five-point scale questionnaire, where 1 means very dissatisfied, 2 means dissatisfied, 3 means average, 4 means relatively satisfied, and 5 means very satisfied. The questionnaire is distributed to employees for investigation, and the scoring data of different employees of different ages and different types of work on the questionnaire are collected;

[0012] Step 5: Data preprocessing: standardize the score data of the observed variables corresponding to the major latent variables collected, remove invalid questionnaires, fill in missing data, and delete low-quality questionnaire data with careless answers and the same answers to all items;

[0013] Step 6: Reliability and validity test: Use SPSS26.0 software to test the reliability and validity of each part of the scoring data after preprocessing in step 5, determine the composition of influencing factors, test the internal reliability of the collected questionnaire data, and verify whether the questionnaire is reliable, stable and effective;

[0014] Step 7: Construct a structural equation model. Use the AMOS 26.0 software to construct a structural equation model of the relationship between the influencing factors of unsafe behaviors based on the data in step 6 that have passed the reliability and validity test and can continue to be analyzed. Use the AMOS 26.0 software to build the measurement model and structural model.

[0015] Step 8: Determine whether the indicators meet the standards. Use AMOS26.0 software to calculate the matching degree of the input data and the constructed model, obtain the fitting degree index, and perform model fitting degree and significance test to determine whether the model meets the adaptation standard conditions;

[0016] Step nine: standardized path output or structural model modification. If the fitting indicators in step eight meet the standard conditions, the standardized path coefficient and significance are obtained; if not, the structural model is modified and adjusted.

[0017] Step 10: Re-test the revised model, and carry out steps 8 and 9 on the revised model until all the fitting indicators of the constructed structural equation model meet the standards. Then, test the revised model to ensure the scientificity and correctness of the model.

[0018] Step 11: Analyze the path. Analyze the path of the model that reaches the quasi-appropriate fit. The relationship hypothesis in the constructed model, that is, the hypothesis proposed in step 2, can be verified by the critical (CR) ratio, standard error (SE), and significant P value between the latent variables. When the significant P value is less than 0.05, it means that the corresponding hypothesis of the path is established, otherwise the hypothesis is not established. According to the sign of the standardized coefficient in step 8, the influence of the influencing factors on the unsafe behavior and the influence relationship between the major factors are judged. When the sign of the standardized coefficient is "+", it means a positive influence; when the sign of the standardized coefficient is "-", it means a negative influence.

[0019] In the above technical solution, the first step is to take the workers of the unit as the research object, observe the workers' behaviors at work, review the accident analysis reports in recent years, interview and other methods, identify and analyze the influencing factors that lead to unsafe behaviors of the workers, where the influencing factors include safety attitude, safety culture, safety education, safety environment, safety ability, safety atmosphere, and unsafe psychology. Taking the workers of the unit as the research object, a variety of methods are adopted to identify and analyze the unsafe behaviors of the workers at work, which improves the reliability of the results obtained and helps to improve the performance and generalization ability of the subsequently established model.

[0020] In the above technical solution, the impact relationship assumption in step 2 is as follows:

[0021] H1: Safety attitude has a significant negative impact on unsafe behavior;

[0022] H2: Safety culture has a significant negative impact on unsafe behaviors;

[0023] H3: Safety education has a significant negative impact on unsafe behavior;

[0024] H4: Safety environment has a significant negative impact on unsafe behavior;

[0025] H5: Safety capability has a significant negative impact on unsafe behavior;

[0026] H6: Safety climate has a significant negative impact on unsafe behaviors;

[0027] H7: Insecurity psychology has a significant positive impact on unsafe behavior;

[0028] H8: Safety attitude and safety capability have a significant impact on each other;

[0029] H9: Safety culture has a significant positive impact on safety attitude;

[0030] H10: Safety culture has a significant positive impact on safety capability;

[0031] H11: Safety climate has a significant positive impact on safety attitude;

[0032] H12: Safety climate has a significant positive impact on safety capability;

[0033] H13: Safety education has a significant positive impact on safety attitude;

[0034] H14: Safety education has a significant positive impact on safety capabilities;

[0035] H15: Insecurity has a significant negative impact on safety attitude;

[0036] H16: Insecurity has a significant negative impact on safety ability;

[0037] H17: The safety environment has a significant negative impact on the unsafe physiology;

[0038] The influence relationship hypothesis provides a theoretical basis for the model, ensuring the scientificity and rationality of model construction.

[0039] In the above technical solution, the latent variables are those variables that cannot be directly measured but can be reflected by observed variables. The latent variables are divided into internal latent variables and external latent variables, and the observed variables are variables whose values ​​can be determined by direct measurement. The observed variables are divided into internal observed variables and external observed variables. There will be a mutual influence relationship between the latent variables. The internal latent variables are variables affected by other latent variables, and the external latent variables are variables affected by other factors outside the system. The external observed variables are determined by external factors and can be directly observed or measured. They are not affected by other variables in the model, but act as input variables to affect other variables in the model. The internal observed variables are determined by other variables inside the model (including external variables and internal variables) and can also be directly observed or measured. In practice, the observed variables are data obtained from the questionnaire.

[0040] In the above technical solution, the reliability and validity test using SPSS26.0 software in step 6 is performed by Cronbach α in reliability analysis.

[0041]

[0042] (n is the number of samples, S i is the variance of the observed sample, S x is the variance of the total sample);

[0043] When Cronbachα coefficient>0.9, it indicates that the reliability of the questionnaire is excellent; between 0.8-0.9, it indicates good reliability; between 0.7-0.8, it indicates acceptable reliability; between 0.6-0.7, it indicates average reliability; below 0.6, it indicates poor reliability. The questionnaire test questions need to be rearranged. Use confirmatory factor analysis KMO value and Bartlett's sphericity test to verify whether the questionnaire items are reasonable. When the KMO value is greater than 0.6 and the Bartlett's sphericity test value is less than 0.05, it indicates that the data continues to be analyzed later.

[0044] In the above technical solution, the structural equation model in step seven includes a measurement model and a structural model;

[0045] The general equation of the measurement model is:

[0046] X=Λ x ξ+δ

[0047] Y=Λ y η+ε

[0048] Where η is the internal latent variable, ξ is the external latent variable, y is the vector of internal observed variables, and x is the vector of external observed variables. δ and ε represent the measurement residuals of the external observed variables and the internal observed variables that are not fully explained by the latent variables, respectively. x is the factor loading matrix of x on ξ, Λ y is the factor loading matrix of y on η;

[0049] Structural equation modeling is a comprehensive statistical analysis method that integrates factor analysis and path analysis. It can study both the internal structural relationship of variables and the relationship between variables.

[0050] The general equation of the structural model is:

[0051] η=Βη+Γξ+ζ

[0052] Where Β is the coefficient matrix, representing the relationship between the internal latent variables η, Γ is the coefficient matrix, representing the impact of the external latent variables on the internal latent variables, and ζ is the estimated error that cannot be fully explained by the internal latent variables.

[0053] In the above technical solution, the parameter results obtained by the AMOS26.0 software in step eight can be estimated by selecting "Calculate estimates" under the menu bar. After the process is completed, select "View text" to view the estimation results of the model parameters;

[0054] The chi-square degrees of freedom (x 2 / df), goodness of fit index (GFI), adjusted goodness of fit index (AGF I), normed fit index (NF I), comparative fit index (CFI), Tucker and Lewis index (TLI), root mean square error of approximation (RMSEA) must all meet standard conditions; when the chi-square degrees of freedom (x 2 When the goodness of fit index (GFI) is greater than 0.8, the adjusted goodness of fit index (AGFI) is greater than 0.8, the normed fit index (NF I) is greater than 0.9, the comparative fit index (CFI) is greater than 0.9, the Tucker and Lewis index (TLI) is greater than 0.9, and the root mean square error of approximation (RMSEA) is less than 0.08, it indicates that the overall fit of the constructed model is good.

[0055] In the above technical solution, the correction methods include deleting the paths where the significance of the external latent variable and the internal latent variable is not obvious, that is, p<0.05, and deleting the paths where the correlation or causal relationship between the external latent variable and the internal latent variable is set unreasonably.

[0056] Beneficial effects of the present invention:

[0057] The invention identifies and analyzes the influencing factors that cause unsafe behaviors of workers through multiple ways, and then assumes the influencing relationship between them. According to the unsafe behavior evaluation index system, the influencing factor indicators are divided into different latent variables and different observed variables corresponding to different latent variables are constructed. The latent variables are used as primary indicators, and the observed variables corresponding to each latent variable are used as secondary indicators. According to the observed variables, a questionnaire is formulated to investigate the employees. The survey data is the observed variable. After preprocessing the survey data, a reliability and validity test is performed. According to the data that pass the reliability and validity test, a structural equation model is constructed, and then the model fit and significance test are performed. If the fit index and significance meet the requirements, the standardized path coefficient and significance are obtained. If not, the model is modified, and then the model fit and significance test is performed again until the fit index and significance meet the requirements. The path of the model that meets the requirements is analyzed, and finally the objective and scientific positive and negative influence relationship between the influencing factors and the positive and negative relationship of the influence of the influencing factors on the unsafe behavior are obtained, so as to provide suggestions for inhibiting the unsafe behavior. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0059] Figure 1A flow chart is established for the structural equation model of the present invention. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0061] like Figure 1 As shown: This embodiment provides an unsafe behavior assessment method based on a structural equation model, comprising the following steps:

[0062] Step 1: Analyze influencing factors, identify and analyze the influencing factors that lead workers to unsafe behaviors;

[0063] Step 2: Propose a hypothesis, a hypothesis about the influence relationship between influencing factors and unsafe behaviors and a hypothesis about the influence relationship between influencing factors;

[0064] Step 3: Divide the indicators. According to the unsafe behavior evaluation index system, divide the influencing factor indicators into different latent variables and construct different observed variables corresponding to different latent variables. Take the latent variables as the primary indicators and the observed variables corresponding to each latent variable as the secondary indicators.

[0065] Step 4: Questionnaire survey. Based on the observed variables corresponding to the different latent variables in step 3, a scale evaluation set of factors affecting unsafe behaviors is designed and compiled, and a questionnaire is constructed. The questionnaire is distributed to employees for investigation. The questionnaire is a Liket five-point scale questionnaire, where 1 means very dissatisfied, 2 means dissatisfied, 3 means average, 4 means relatively satisfied, and 5 means very satisfied. The questionnaire is distributed to employees for investigation, and the scoring data of different employees of different ages and different types of work on the questionnaire are collected;

[0066] Step 5: Data preprocessing: standardize the score data of the observed variables corresponding to the major latent variables collected, remove invalid questionnaires, fill in missing data, and delete low-quality questionnaire data with careless answers and the same answers to all items;

[0067] Step 6: Reliability and validity test: Use SPSS26.0 software to test the reliability and validity of each part of the scoring data after preprocessing in step 5, determine the composition of influencing factors, test the internal reliability of the collected questionnaire data, and verify whether the questionnaire is reliable, stable and effective;

[0068] Step 7: Construct a structural equation model. Use the AMOS 26.0 software to construct a structural equation model of the relationship between the influencing factors of unsafe behaviors based on the data in step 6 that have passed the reliability and validity test and can continue to be analyzed. Use the AMOS 26.0 software to build the measurement model and structural model.

[0069] Step 8: Use AMOS26.0 software to calculate the matching degree between the input data and the constructed model, obtain the matching degree index, and perform model matching and significance tests to determine whether the model meets the adaptation standard conditions;

[0070] Step nine: If the fitting indicators in step eight meet the standard conditions, the standardized path coefficient and significance are obtained; if not, the structural model is modified and adjusted;

[0071] Step 10: Perform steps 8 and 9 on the revised model until all the fitting indicators of the constructed structural equation model meet the standards, and then test the revised model to ensure the scientificity and correctness of the model;

[0072] Step 11: Analyze the path of the model that reaches the quasi-appropriate fit. The relationship hypothesis in the constructed model, that is, the hypothesis proposed in step 2, can be verified by the critical (CR) ratio, standard error (SE), and significant P value between the latent variables. When the significant P value is less than 0.05, it means that the corresponding hypothesis of the path is established, otherwise the hypothesis is not established. According to the sign of the standardized coefficient in step 8, the influence of the influencing factors on the unsafe behavior and the influence relationship between the major factors are judged. When the sign of the standardized coefficient is "+", it means that it is a positive influence; when the sign of the standardized coefficient is "-", it is a negative influence. According to the above steps, an objective and scientific structural equation model for evaluating unsafe behavior can be created, and the positive and negative influence relationship between the influencing factors and the positive and negative relationship between the influencing factors on the unsafe behavior can be intuitively obtained, providing suggestions for suppressing unsafe behavior.

[0073] As an implementation mode of the present invention, the step one is to take the workers of the unit as the research object, observe the workers' behaviors at work, review the accident analysis reports in recent years, interview and other methods, identify and analyze the influencing factors that lead to the workers' unsafe behaviors, where the influencing factors include safety attitude, safety culture, safety education, safety environment, safety ability, safety atmosphere, and unsafe psychology. Taking the workers of the unit as the research object, a variety of methods are adopted to identify and analyze the unsafe behaviors of the workers at work, which improves the reliability of the results obtained and helps to improve the performance and generalization ability of the subsequently established model.

[0074] As an implementation mode of the present invention, the influence relationship assumption in step 2 is as follows:

[0075] H1: Safety attitude has a significant negative impact on unsafe behavior;

[0076] H2: Safety culture has a significant negative impact on unsafe behaviors;

[0077] H3: Safety education has a significant negative impact on unsafe behavior;

[0078] H4: Safety environment has a significant negative impact on unsafe behavior;

[0079] H5: Safety capability has a significant negative impact on unsafe behavior;

[0080] H6: Safety climate has a significant negative impact on unsafe behaviors;

[0081] H7: Insecurity psychology has a significant positive impact on unsafe behavior;

[0082] H8: Safety attitude and safety capability have a significant impact on each other;

[0083] H9: Safety culture has a significant positive impact on safety attitude;

[0084] H10: Safety culture has a significant positive impact on safety capability;

[0085] H11: Safety climate has a significant positive impact on safety attitude;

[0086] H12: Safety climate has a significant positive impact on safety capability;

[0087] H13: Safety education has a significant positive impact on safety attitude;

[0088] H14: Safety education has a significant positive impact on safety capabilities;

[0089] H15: Insecurity has a significant negative impact on safety attitude;

[0090] H16: Insecurity has a significant negative impact on safety ability;

[0091] H17: The safety environment has a significant negative impact on the unsafe physiology;

[0092] The influence relationship hypothesis clarifies the relationship between influencing factors and between influencing factors and unsafe behaviors, guides the structural design of the model, provides a theoretical basis for the model, and ensures the scientificity and rationality of model construction.

[0093] As an implementation mode of the present invention, the latent variables are those variables that cannot be directly measured but can be reflected by observed variables. The latent variables are divided into internal latent variables and external latent variables, and the observed variables are variables that can be determined by direct measurement. The observed variables are divided into internal observed variables and external observed variables. There will be a mutual influence relationship between the latent variables. The internal latent variables are variables affected by other latent variables, and the external latent variables are variables affected by other factors outside the system. The external observed variables are variables determined by external factors and can be directly observed or measured. They are not affected by other variables in the model, but act as input variables to affect other variables in the model. The internal observed variables are determined by other variables (including external variables and internal variables) inside the model, and can also be directly observed or measured. In practice, the observed variables are data obtained from the questionnaire, and the questionnaire adopts Li The questionnaire is in the form of a five-point scale with high reliability and validity. The same variable can be measured through multiple items. The measurement results have high reliability and validity, which is convenient for statistical analysis and the use of complex statistical tools to conduct in-depth analysis of the survey results. It is easy to operate and the questionnaire design is relatively simple, which is easy for respondents to understand and answer. It is also convenient for researchers to collect and process data.

[0094] As an embodiment of the present invention, the reliability and validity test using SPSS26.0 software in step 6 is performed by Cronbach α in reliability analysis.

[0095]

[0096] (n is the number of samples, S i is the variance of the observed sample, S x is the variance of the total sample);

[0097] When Cronbachα coefficient>0.9, it means that the questionnaire has excellent reliability, 0.8-0.9 indicates good reliability, 0.7-0.8 indicates acceptable reliability, 0.6-0.7 indicates average reliability, and below 0.6 indicates poor reliability. The questionnaire test questions need to be rearranged. Use the confirmatory factor analysis KMO value and Bartlett's sphericity test to verify whether the items in the questionnaire are reasonable. When the KMO value is greater than 0.6 and the Bartlett's sphericity test value is less than 0.05, the data will continue to be analyzed. The calculation of the Cronbach coefficient takes into account the variance of each item in the questionnaire or scale and the covariance between the items. By comparing these variances and covariances, a coefficient reflecting internal consistency can be obtained to judge the accuracy and reliability of the measurement results.

[0098] As an implementation mode of the present invention, the structural equation model in step seven includes a measurement model and a structural model;

[0099] The general equation of the measurement model is:

[0100] X=Λ x ξ+δ

[0101] Y=Λ y η+ε

[0102] Where η is the internal latent variable, ξ is the external latent variable, y is the vector of internal observed variables, and x is the vector of external observed variables. δ and ε represent the measurement residuals of the external observed variables and the internal observed variables that are not fully explained by the latent variables, respectively. x is the factor loading matrix of x on ξ, Λ y is the factor loading matrix of y on η;

[0103] Structural equation modeling is a comprehensive statistical analysis method that integrates factor analysis and path analysis. It can study both the internal structural relationship of variables and the relationship between variables.

[0104] The general equation of the structural model is:

[0105] η=Βη+Γξ+ζ

[0106] Where Β is the coefficient matrix, representing the relationship between the endogenous latent variables η; Γ is the coefficient matrix, representing the impact of the exogenous latent variables on the endogenous latent variables; ζ is the estimated error that cannot be fully explained by the endogenous latent variables; the measurement model focuses on the relationship between the observed variables and the latent variables; the structural model focuses on the relationship between the latent variables, allowing researchers to test the hypothesized causal relationship, which can be linear or nonlinear; the structural equation model is a statistical method based on the covariance matrix of the variables to analyze the relationship between the variables, integrating factor analysis and path analysis, and can test the relationship between the observed variables, latent variables and error variables in the model, thereby obtaining the direct effect, indirect effect and total effect of the independent variable on the dependent variable.

[0107] As an implementation mode of the present invention, the various parameter results obtained by the AMOS26.0 software in step eight can be estimated by selecting "Calculate estimates" in the menu bar. After the process is completed, select "View text" to view the estimated results of various model parameters, where the chi-square degrees of freedom (x 2 / df), goodness of fit index (GF I), adjusted goodness of fit index (AGF I), normed fit index (NF I), comparative fit index (CF I), Tucker and Lewis index (TL I), root mean square error of approximation (RMSEA) must all meet standard conditions; when the chi-square degrees of freedom (x 2 / df) is less than 3, the goodness of fit index (GF I) is greater than 0.8, the adjusted goodness of fit index (AGF I) is greater than 0.8, the normed fit index (NF I) is greater than 0.9, the comparative fit index (CF I) is greater than 0.9, the Tucker and Lewis index (TL I) is greater than 0.9, and the root mean square error of approximation (RMSEA) is less than 0.08, which indicates that the overall fit of the constructed model is good.

[0108] As an implementation mode of the present invention, the correction method includes deleting the path where the significance of the external latent variable and the internal latent variable is not obvious, that is, p<0.05, deleting the path where the correlation or causal relationship between the external latent variable and the internal latent variable is set unreasonable, and after the correction, the model fit and significance test are re-performed until the fit index and significance meet the requirements, and the path of the model that meets the requirements is analyzed, and finally, the objective and scientific positive and negative influence relationship between the influencing factors and the positive and negative influence relationship of the influencing factors on the unsafe behavior are obtained, so as to provide suggestions for suppressing the unsafe behavior.

[0109] It can be understood that this paper identifies and analyzes the influencing factors that cause workers to engage in unsafe behaviors and hypothesizes the influence relationship between them, then divides them into different latent variables and constructs their corresponding observed variables, formulates a questionnaire to survey employees, constructs a structural equation model based on the data that pass the reliability and validity test, and then conducts a test of the model's fit and significance. If the requirements are met, the standardized path coefficient and significance are obtained. If not, the model is modified, and then the model's fit and significance are retested until the requirements are met. The path of the model that meets the requirements is analyzed, and finally, an objective and scientific positive and negative influence relationship between the influencing factors and the positive and negative relationship between the influencing factors and the unsafe behavior are obtained, providing suggestions for curbing unsafe behaviors.

[0110] In the description of the present invention, unless otherwise specified, "plurality" means two or more than two; it should be understood that the terms "opening", "upper", "lower", "thickness", "top", "middle", "length", "inside", "around" and the like indicate orientation or positional relationship, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the components or elements referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention.

[0111] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for evaluating unsafe behavior based on a structural equation model, characterized in that: The steps include: Step 1: Analyze influencing factors, identify and analyze the influencing factors that lead workers to unsafe behaviors; Step 2: Propose a hypothesis, a hypothesis about the influence relationship between influencing factors and unsafe behaviors and a hypothesis about the influence relationship between influencing factors; Step 3: Divide the indicators. According to the unsafe behavior evaluation index system, divide the influencing factor indicators into different latent variables and construct different observed variables corresponding to different latent variables. Take the latent variables as the primary indicators and the observed variables corresponding to each latent variable as the secondary indicators. Step 4: Questionnaire survey: design and compile a scale evaluation set of factors affecting unsafe behaviors based on the observed variables corresponding to the different latent variables in step 3 and construct a questionnaire, which is then distributed to employees for investigation; Step 5: Data preprocessing: standardize and preprocess the score data of the observed variables corresponding to the major latent variables collected; Step 6: Reliability and validity test: Use SPSS26.0 software to test the reliability and validity of each part of the scoring data after preprocessing in step 5 to determine the composition of influencing factors; Step 7: Construct a structural equation model. Use the AMOS 26.0 software to construct a structural equation model of the relationship between the influencing factors of unsafe behaviors based on the data in step 6 that can be used for subsequent analysis after reliability and validity tests. Step 8: Determine whether the indicators meet the standards. Use AMOS26.0 software to calculate the matching degree between the input data and the constructed model, and determine whether each fitting indicator meets the adaptation standard conditions; Step nine: standardized path output or structural model modification. If the fitting indicators in step eight meet the standard conditions, the standardized path coefficient and significance are obtained; if not, the structural model is modified and adjusted. Step 10: retest the revised model, and perform steps 8 and 9 on the revised model until all the fit indicators of the constructed structural equation model meet the standards; Step 11, path analysis, analyzes the path to the model that reaches the quasi-fit. The relationship hypothesis in the constructed model, that is, the hypothesis proposed in step 2, can be verified by the critical (CR) ratio, standard error (SE), and significant P value between each latent variable. When the significant P value is less than 0.05, it means that the corresponding hypothesis of the path is established, otherwise the hypothesis is not established.

2. The unsafe behavior assessment method based on the structural equation model according to claim 1 is characterized by: The first step is to take the workers of the unit as the research object, observe the workers' behavior at work, review the accident analysis reports in recent years, interview and other methods, identify and analyze the influencing factors that lead to the workers' unsafe behavior, among which the influencing factors include safety attitude, safety culture, safety education, safety environment, safety ability, safety atmosphere, and unsafe psychology.

3. The unsafe behavior assessment method based on the structural equation model according to claim 2 is characterized by: The impact relationship assumptions in step 2 are as follows: H1: Safety attitude has a significant negative impact on unsafe behavior; H2: Safety culture has a significant negative impact on unsafe behaviors; H3: Safety education has a significant negative impact on unsafe behavior; H4: Safety environment has a significant negative impact on unsafe behavior; H5: Safety capability has a significant negative impact on unsafe behavior; H6: Safety climate has a significant negative impact on unsafe behaviors; H7: Insecurity psychology has a significant positive impact on unsafe behavior; H8: Safety attitude and safety capability have a significant impact on each other; H9: Safety culture has a significant positive impact on safety attitude; H10: Safety culture has a significant positive impact on safety capability; H11: Safety climate has a significant positive impact on safety attitude; H12: Safety climate has a significant positive impact on safety capability; H13: Safety education has a significant positive impact on safety attitude; H14: Safety education has a significant positive impact on safety capabilities; H15: Insecurity has a significant negative impact on safety attitude; H16: Insecurity has a significant negative impact on safety capabilities; H17: Safety environment has a significant negative impact on insecurity physiology.

4. The unsafe behavior assessment method based on the structural equation model according to claim 1 is characterized by: The latent variables are those variables that cannot be directly measured but can be reflected by observed variables. The latent variables are divided into internal latent variables and external latent variables. The observed variables are variables that can be determined by direct measurement. The observed variables are divided into internal observed variables and external observed variables.

5. The unsafe behavior assessment method based on the structural equation model according to claim 1 is characterized by: The reliability and validity test using SPSS26.0 software in step 6 is carried out through the Cronbach α coefficient in the reliability analysis. (n is the number of samples, S i is the variance of the observed sample, S x is the variance of the total sample) When Cronbach α is > 0.9, it indicates that the reliability of the questionnaire is excellent; when it is between 0.8 and 0.9, it indicates that the reliability is good; when it is between 0.7 and 0.8, it indicates that the reliability is acceptable; when it is between 0.6 and 0.7, it indicates that the reliability is average; when it is lower than 0.6, it indicates that the reliability is poor and the questionnaire test questions need to be rearranged; The confirmatory factor analysis KMO value and Bartlett's sphericity test were used to verify whether the items in the questionnaire were reasonable. When the KMO value was greater than 0.6 and the Bartlett's sphericity test value was less than 0.05, it indicated that the data continued to be analyzed later.

6. The unsafe behavior assessment method based on structural equation model according to claim 1 is characterized by: The structural equation model in step seven includes a measurement model and a structural model; The general equation of the measurement model is: X=Λ x ξ+d Y=Λ y the+e Where η is the internal latent variable, ξ is the external latent variable, y is the vector of internal observed variables, and x is the vector of external observed variables. δ and ε represent the measurement residuals of the external observed variables and the internal observed variables that are not fully explained by the latent variables, respectively. x is the factor loading matrix of x on ξ, Λ y is the factor loading matrix of y on η; Structural equation modeling is a comprehensive statistical analysis method that integrates factor analysis and path analysis. It can study both the internal structural relationship of variables and the relationship between variables. The general equation of the structural model is: η=Βη+Γξ+ζ Where Β is the coefficient matrix, representing the relationship between the internal latent variables η, Γ is the coefficient matrix, representing the impact of the external latent variables on the internal latent variables, and ζ is the estimated error that cannot be fully explained by the internal latent variables.

7. The unsafe behavior assessment method based on structural equation model according to claim 1 is characterized by: The parameter results obtained by the AMOS26.0 software in step 8 can be estimated by selecting "Calculateestimates" in the menu bar. After the process is completed, select "View text" to view the estimated results of the model parameters; The chi-square degrees of freedom (x 2 / df), goodness of fit index (GFI), adjusted goodness of fit index (AGFI), normed fit index (NFI), comparative fit index (CFI), Tucker and Lewis index (TLI), root mean square error of approximation (RMSEA) must all meet standard conditions; when the chi-square degrees of freedom (x 2 / df) is less than 3, the goodness of fit index (GFI) is greater than 0.8, the adjusted goodness of fit index (AGFI) is greater than 0.8, the normed fit index (NFI) is greater than 0.9, the comparative fit index (CFI) is greater than 0.9, the Tucker and Lewis index (TLI) is greater than 0.9, and the root mean square error of approximation (RMSEA) is less than 0.08, which indicates that the overall fit of the constructed model is good.

8. The unsafe behavior assessment method based on structural equation model according to claim 1 is characterized by: The correction methods include deleting the paths where the significance of the external latent variable and the internal latent variable is not obvious, that is, p<0.05, and deleting the paths where the correlation or causal relationship between the external latent variable and the internal latent variable is unreasonable.

Images