Safety disconnect: Analysis of the role of labor experience and safety training on work safety perceptions

Abstract

Purpose: This study analyzes work safety perceptions among workers and safety experts in the construction industry. Furthermore, we evaluate whether experiential learning—i.e., labor experience—and knowledge-enhancing practices—that we link to safety training—explain the differences in work safety perceptions of workers and safety experts by triggering different types of overconfidence biases. Design/methodology/approach: The proposed hypothesis are tested by applying ordered probit models on a unique dataset comprising information for 558 employees and 215 safety experts working in the Spanish construction sector. Findings: The results reveal that previous labor market experience has a significantly negative effect on perceived work safety, that is, risk awareness decreases with respect to labor experience. However, the findings indicate that differences in perceived work safety between workers and safety experts are not explained by previous labor experience. Furthermore, the results suggest that higher levels of safety training—which we link to the acquisition of codified knowledge—negatively impacts workers’ safety perceptions, while this effect turns positive among safety experts. This result suggests that safety experts’ perceived work safety is affected by overconfidence that results from their greater safety-specific training (over-precision bias). Originality/value: Work safety constitutes a relevant key performance indicator. The proposed analysis of the role of labor experience and safety training on perceived work safety in different types of employees contributes to better understand how organizations can improve the management of their workforce by triggering specific actions—such as the design of customized training programs—that may help in reducing the safety disconnect between employees, in terms of perceived work safety.

Keywords

• Safety disconnect, perceived work safety, overconfidence, over-estimation bias, over-precision bias, ordered probit model, workforce management