Dynamic Occupational Accidents Modeling Using Dynamic Hybrid Bayesian Confirmatory Factor Analysis: An In-Depth Psychometrics Study Publisher



Zarei E¹ ; Karimi A² ; Habibi E³ ; Barkhordari A² ; Reniers G⁴ Show Affiliations
Source: Safety Science Published:2021

Abstract

Multiple factors contribute to occupational accidents including individual, job, environmental, organizational, and family issues. Most of them are latent factors and hard to model how and what extent they influence incidents occurrence. Although past research has included an occupational incident context, this attention has rarely provided a holistic instrument for the dynamic causal modeling of different influencing factors. Hence, the present study is aimed at developing a concrete instrument for identifying and modeling various influencing factors on occupational accident occurrence. After a comprehensive literature review and employing occupational safety and industrial psychological experts to achieve a reasonable conceptual model, the primary structure of the instrument was developed. Several systematic attempts were made to identify items (questions), define contributing factors, assess content and face validity, analyze its reliability, construct validity, criterion validity, and assess the model's fitness using advanced statistics tests by SPSS.22 and LISREL9.3 programs. Finally, dynamic hybrid Bayesian Network (DHBN) based on the Confirmatory Factor Analysis (CFA) technique was developed to model accident occurrence and simulate the behavior of the main influencing factors over ten years under uncertainty. After standardization of the proposed instrument, a comprehensive study was conducted via the participation of 700 workers from thirty-eight manufacturing companies to illustrate its effectiveness and modeling capability. The findings revealed the effectiveness of the proposed instrument in the causation modeling of occupational incidents, dynamic modeling of contributing factors, and risk-based decision making for occupational incidents management. The proposed instrument can serve as a holistic tool for accurately identifying and dynamic modeling of different latent influencing factors, and making tailored safety decisions in different workplace context. © 2020 Elsevier Ltd