Why do Workers Generate Biased Risk Perceptions? An Analysis of Anchoring Effects and Influential Factors in Workers' Assessment of Unsafe Behavior

Abstract

Background

Risk perception plays a crucial role in workers' unsafe behaviors. However, little research has explored why workers generate biased risk perceptions, namely underestimating or overestimating the risks of unsafe actions. Cognitive biases in risk perception arise from uncertainties about the dangers of unsafe behaviors. As a typical heuristic strategy, the anchoring effect is critical in decision-making under uncertain conditions. Consequently, this study empirically analyzed the influence of anchoring effects on workers' risk perception.

Methods

In 2022, a survey was conducted with 1,418 coal mine workers from Shanxi Province, China. The survey instruments assessed workers' risk perception of unsafe behavior, anchoring effects, need for cognition, and safety knowledge. Multivariable linear regression models were employed to analyze the associations among these variables.

Results

The findings verified the proposed anchoring effects. Specifically, experimenter-provided high-risk anchors led workers to overestimate unsafe behavior risks, thus reducing their tendency to engage in such behavior. In contrast, experimenter-provided low-risk anchors and accident-injury experiences (self-generated anchors) decreased workers' risk perception, increasing their propensity to engage in unsafe behavior. Additionally, workers' safety knowledge and need for cognition significantly affected anchoring effects.

Conclusion

This research enhances workplace safety studies by applying the anchoring effect from psychology to risk perception research. Suggestions for improving risk perception encompass implementing hazard warnings, fostering safety education, and providing training. Furthermore, managers should give special attention to workers with accident-injury experience and promptly correct their accident fluke mentality, thereby improving overall risk awareness.