A fuzzy Bayesian network DEMATEL model for predicting safety behavior

Abstract

Objectives. Safety behavior significantly affects safety performance in the workplace. This study aimed to develop a Bayesian network (BN) model for managing and improving safety behavior. Methods. This study was carried out in the chemical industries in Iran. The data were gathered by a questionnaire consisting of 13 variables including organization safety priority, systems design, safety communication, safety education, work strategy, human–system interaction, mental workload, environmental distractions, work pressure, fatigue, sleepiness, safety knowledge and locus of control. The BN structure was created using the fuzzy decision-making trial, evaluation laboratory method and expert opinions. Belief updating was used to determine variables with the strongest effect on safety behavior. Results. Locus of control, organization safety priority and safety knowledge were the best predictors of safety behavior. Moreover, it was found that improving organization safety priority and safety knowledge is the best intervention strategy to improve safety behavior significantly. Conclusions. BN is a powerful tool that can model causal relationships among variables. Improving organization safety priority and safety knowledge can lead to the maximum possible level of safety behavior.

Keywords:

• Bayesian network
• fuzzy DEMATEL
• safety behavior
• safety knowledge

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the Qom University of Medical Sciences [grant number 991221].