Constructing a Bayesian network model for improving safety behavior of employees at workplaces

Area of Science:

• Occupational Health and Safety
• Risk Management
• Behavioral Science

Background:

• Unsafe workplace behaviors significantly elevate accident risks, necessitating effective management strategies.
• Existing models for explaining and improving safety behavior are often inadequate.
• This research addresses the need for a robust framework to manage and enhance employee safety practices.

Purpose of the Study:

• To develop a model for managing and improving employee safety behavior using the Bayesian networks approach.
• To identify key factors influencing safety behavior in power plant construction.
• To determine optimal intervention strategies for enhancing workplace safety.

Main Methods:

• Data collection via questionnaires on nine factors (e.g., management commitment, safety knowledge, work pressure) in Iranian power plant construction projects.
• Construction of a Bayesian network model incorporating expert opinions and Dempster-Shafer theory.
• Application of belief updating to identify and select the most effective safety improvement strategies.

Main Results:

• A majority of employees exhibit a tendency to disregard safety rules and procedures.
• Safety attitude, safety knowledge, and a supportive environment were identified as the strongest predictors of safety behavior.
• Enhancing the supporting environment and employee participation emerged as the most effective strategy for immediate safety improvements.

Conclusions:

• Bayesian networks offer a valuable tool for analyzing and improving workplace safety behavior.
• The belief updating feature of Bayesian networks facilitates the comparison and selection of intervention strategies.
• A comprehensive model is needed to fully explain and manage safety behavior effectively.