Beyond human error: An empirical study of the safety Model 1 and Model 2 approaches for predicting workers' behaviors and outcomes with procedures. (February 2021)

• Record Type: Journal Article
• Authors: Hendricks, Joseph W.; Peres, S. Camille
• Is Part Of: Safety science. Volume 134(2021)
• Publication Date: 2021-02
• Subjects: Procedures -- Process safety -- Models of safety -- Predictors of non-compliance; Industrial accidents -- Periodicals; Accident Prevention -- Periodicals; Safety -- Periodicals; Travail -- Accidents -- Périodiques; 363.11
• DOI: 10.1016/j.ssci.2020.105016
• Languages: English
• ISSNs: 0925-7535

Abstract

Highlights: Study of system- and individual-level variables with incidents and behaviors with procedure. Procedure quality and usefulness matter for predicting incidents and behaviors with procedures. Safety Model 1 and 2 needed to predict humans' interactions in complex socio-technical systems. Abstract: Procedures serve an important role as safety barriers in industries and can assist workers in executing tasks. However, several notable incidents have occurred, such as the Texas City Refinery and Piper Alpha disasters, where the lack of compliance with procedures was a contributing cause. Historically, much of the research and practice has focused on better controlling the workers' behaviors and increasing compliance through strict adherence to written procedures. Dekker (2003) refers to this as a Model 1 approach and compares it to a Model 2 approach where procedures are designed to be resources or tools to support workers' actions while performing a task. While the Model 1 posits that individual-level variables are more predictive of important outcomes, Model 2 suggests that individual and system-level variables are both important. The current empirical study ( N = 174) sought to examine both system- and individual-level variables associated with behaviors and outcomes in procedural systems to support the need to include Model 2 in procedural systems research and practice. Results indicated that attitudinal individual-level variables regarding procedure utility and compliance uniquely predicted relevant criteria (deviations, use, and incident-near-miss rates) in multiple regression models. Additionally, we found that system-level variables did improve two out of three models, and procedure quality perceptions were a significant system-level predictor in all of the full regression models. These results illustrate the importance of moving away from an exclusive Model 1 approach to safety and procedures in high-risk, industrial organizations.